Generation of CD8⁺ Cytotoxic T Cell Clones Recognizing BMI1-Derived Peptides

Leukemic stem cells (LSCs) resistant to conventional chemotherapeutic drugs can cause relapse of disease even after intensive treatment. Therefore, novel therapeutic strategies targeting LSCs are needed. The polycomb group gene BMI1 was initially identified as a cooperating oncogene in Eμ/myc transgenic mice. The protein product of BMI1 has been shown to repress the products of the CDKN2 tumor suppressor gene, p16 (INK4A) and p14 (ARF), and is required for self-renewal of both normal hematopoietic stem cells as well as LSCs. BMI1 expression level has been correlated with prognosis in myelodysplastic syndrome (MDS), acute myeloid leukemia (AML), and chronic myeloid leukemia (CML). Furthermore, higher BMI1 expression in the tumorigenic cancer stem cell population has been demonstrated in brain tumor, breast cancer, and head and neck cancer. Thus, BMI1 represents a potential attractive target for cancer stem cell-specific immunotherapy. To further evaluate BMI1 potential, we generated CD8⁺BMI1-specific cytotoxic T cell clones. Two BMI1-encoded peptides (CLPSPSTPV, BMI1_271-279, and TLQDIVYKL, BMI1_74-82) with predicted high affinity binding to the class I MHC molecule HLA-A*0201 were identified using publicly available prediction algorithms, and high-affinity binding to HLA-A*0201 was confirmed using an MHC stabilization assay on T2 cells. CD8⁺ T cells were negatively enriched from peripheral blood mononuclear cells (PBMC) from three normal human HLA-A 0201⁺ donors and stimulated at 10-day intervals with BMI1 peptide-pulsed autologous dendritic cells or PBMC for two or three cycles. Reactivity of the resulting T cell lines for the corresponding BMI1 peptides was subsequently assessed by ELISPOT assay, and was detected in lines from two of the three donors for CLPSPSTPV and in a line from one of the three donors for TLQDIVYKL. A CTL clone specific for CLPSPSTPV was isolated from one of the donors and evaluated in further studies. A peptide titration assay demonstrated that the clone had high affinity (peptide concentration required for half maximal lysis at E/T=10:1, 0.1 nM). In standard cytotoxicity assays, this CTL clone efficiently killed CLPSPSTPV peptide-pulsed T2 cells and 6 of 9 HLA-A*0201-positive or -transduced BMI1⁺ myeloid leukemia cell lines, and did not kill nonmalignant HLA-A2⁺ cells, including fibroblasts, PHA-blasts, and EBV-transformed B cells. Quantitative RT-PCR analysis of BMI1 expression was performed on a commercial multiple tissue cDNA panel, 39 primary leukemia/lymphoma samples (4 ALL, 22 AML, 4 CLL, 4 DLBCL, 4 FL, 1 MCL), and 31 solid tumor lines (21 Renal cell ca., 1 Ovarian ca., 3 Breast ca., 4 Melanoma, 1 Colon ca., 1 Pancreas ca.). All samples had detectable levels of BMI1 expression. Among normal tissues, testis had the highest level of BMI1 expression. Of the primary leukemia/lymphoma samples, four of 22 AML cells and one of four CLL cells expressed high levels of BMI1, and all solid tumor lines had low levels of BMI1 expression. When tested for cytotoxicity against five HLA-A*0201⁺ primary leukemia cells (3 AML, 1 ALL, 1 CML-BC), the BMI1-specific CTL clone killed 3 of 5 efficiently. The clone was also tested for recognition of 14 HLA-A*0201⁺ solid tumor lines (6 Renal cell ca., 2 Breast ca. 2 Melanoma, 1 Ovarian ca., 1 Colon ca., 2 Medulloblastoma), but only killed 2 of the 14 lines. These results suggest that CTL therapy targeting BMI1 may have utility as a novel strategy for eliminating LSCs. Future studies will determine whether this clone can specifically recognize leukemic stem cells and inhibit the engraftment of human AML in NOD/SCID mice.