A CD8+ T-Cell Clone Directed Against AML Blasts that Recognizes a Tumor Specific Antigen Expressed Also by Solid Tumors

Abstract 4289

The aim of this study was to generate cytotoxic T-lymphocytes (CTL) clones directed against AML cells and to identify new immunogenic antigens with the following properties: i) to be overexpressed by leukemic cells and not by normal tissues; ii) to be shared among different AML subtypes; iii) to play a role in leukemic growth/survival; iv) to be expressed by leukemic stem cells.

To this end, we loaded normal dendritic cells (DC) from a healthy donor with apoptotic bodies from primary AML cells and used loaded DC to stimulate autologous lymphocytes (i.e. lymphocytes from the donor). Donor was selected to be partially matched with the leukemic patient for MHC-class I, in that he shared two MHC-class I alleles at the HLA-supertype level (HLA-B7 and HLA-B44). With this strategy, a CD8+ T-cell line was generated that recognized both loaded DC and AML cells used for loading, but not DC loaded with normal cells derived from the patient (PHA-blasts). This CTL line was cloned by limiting dilutions and 180 clones were screened by IFN-g elispot against 2 different AML samples that were expressing both HLA-B7 and HLA-B44; one additional mismatched AML sample was used as negative control. Two clones were selected that recognized HLA-B7+/HLA-B44+ AML cells but not the negative control (namely, clone 31D3 and 8E12). To determine the HLA-restriction element and to verify shared antigen expression among AML subtypes, we tested both clones against a panel of 18 HLA-typed primary AML samples. We found that clone 31D3 was restricted by HLA-B7 and clone 8E12 by HLA-B44. Each clone recognized 5/5 HLA-matched AML samples of different subtype. In addition, the clones did not recognize both resting and activated normal myeloid, nor lymphoid and CD34+ cells expressing the proper HLA-restriction allele. In addition to elispot activity, both clones showed killing of AML cells in a CFSE-based cytotoxic assay. To confirm the lack of reactivity against normal tissues, we analyzed the activity of both clones against HLA matched or mismatched fibroblasts and we found that clone 8E12 displayed a low reactivity activity against normal matched fibroblasts, while clone 31D3 was not reactive.

We then focused the analysis on clone 31D3 and tested whether its activity was restricted to leukemia or also directed against a panel of HLA-B7 positive (N=3) or negative (N=5) solid tumors. Interestingly, we found that the clone 31D3 could recognize one colon carcinoma and one melanoma cell line expressing the HLA-B7 supertype. In particular, the melanoma cell line (G4-mel) was HLA-B35+ (HLA-B35 belongs to the HLA-B7 supertype) and was recognized at very high levels. The availability of a tumor cell line expressing adequate amounts of antigen will make the generation of a cDNA library for antigen identification more feasible than with primary leukemic cells. Finally, to further confirm that HLA-B35 is the proper MHC restriction element and to determine to what extent the antigen is shared among different tumors, we transduced 8 tumor cell lines, 3 normal B-lymphoblastoid and 1 fibroblast cell line with the HLA-B35 allele. Clone 31D3 efficiently recognized 6/8 HLA-B35 transduced tumor cell lines (1 AML, 2 melanomas, 2 colon and 1 breast carcinoma) but none of the control cell lines.

In conclusion, clone 31D3 is restricted by the HLA-B7 supertype, it recognizes an antigen that is expressed by leukemic cells and not by normal tissues and, most importantly, is shared among tumors of different histology. Since the vast majority of malignant cells tested (both primary cells and in vitro established cell lines) were targeted by this clone, the putative antigen might be a protein playing an important role in tumor growth or survival. We are now testing the activity of the clone against purified leukemic stem cells and attempting to identify this shared tumor antigen.