Generation of Multi-Antigen Specific T Cells for Adoptive Immunotherapy of Myeloid Leukemia and Identification of MHC Class I and II-Restricted Peptides for WT1, Proteinase 3 and Human Neutrophil Elastase

Abstract 2985

Generating cytotoxic T lymphocytes (CTL) against malignant cells using defined leukemia associated antigen (LAA) vaccines, or ex vivo adoptively transferred T cells, is a promising approach to prevent or treat leukemia. Vaccine or T cell therapies have largely focused on CD8 responses to defined HLA-A2 restricted LAA peptides. Strategies targeting single peptides are limited to the recipient's HLA type, lack of CD4 help, and risk of tumor immune escape. To overcome these constraints and develop T cell immunotherapy for patients with myeloid malignancies, we used entire peptide libraries of several LAA to generate multi-specific CD4+ and CD8+ leukemia–reactive CTL. We obtained 15mer peptide libraries overlapping by 11 aa for WT1, Proteinase 3 (Pr3), MAGE-A3, PRAME and human neutrophil elastase (HNE). PBMC of 10 healthy donors were stimulated with DC pulsed with peptide libraries of one antigen or the mix of 5 LAA (LAA mix). Cultures were re-stimulated weekly with peptide-pulsed DCs and peptide-recognition was tested by IFN γ-ELIspot. Positive cultures were further characterized by performing intracellular cytokine staining, measuring degranulation by CD107a release, and assessing in vitro cytolytic activity using peptide-pulsed autologous PHA blasts as targets in a flow cytometric killing assay.

Nine out of 10 healthy donors responded to the LAA mix as assessed by IFN γ-ELIspot (mean: 471.2 SFU/2×10⁵ cells, range: 296–968.5). The expanded CTL specifically recognized 3 antigens (mean, range 1–5), and PRAME was most frequently recognized (in 9 of 10 donors screened). Within donors we consistently saw recognition of the same profile of antigens irrespective of whether the CTL were induced with individual antigen or LAA mix, indicating that antigenic competition between LAA was not a limiting factor in our cultures. Correlating with ELIspot results, LAA CTL were cytotoxic to PHAB pulsed with the LAA mix (E:T=40:1; mean: 33.1%, range: 12.4–80.0%; negative control: PHAB: mean: 4.2%, range: 0–13.4%; n=7), as well as individual antigens (WT1: mean :12.2%, range: 0–26.1%; Pr3: mean: 9.7%, range: 5.6–13.7%; HNE: mean: 35.4%, range: 0–84.9%; MAGE-A3: mean: 19.9%, range: 15.2–28.2%; PRAME: mean: 16.7%, range: 3.8–37.5%). Cytotoxicity against the peptide library or the LAA mix was greater than against targets pulsed with the immunodominant peptide, demonstrating the benefit of a broad immune response. Cytolytic activity was also seen in donors with predominantly CD4+ response, suggesting a role for T helper cells in tumor lysis.

The approach of stimulation with LAA mix was also feasible for epitope mapping. In addition to known peptides, we identified 16 new MHC class I and II-restricted peptides (WT1: 8, HNE: 6 and Pr3: 2). Thereby, up to 4 immunogenic peptides of two different antigens could be identified in one LAA mix culture. WT1 peptides showed class I and class II-restriction by blocking experiments with HLA antibodies. HNE peptides were mainly class I-restricted, with a dominant response against RRLACLFLACVLPAL in 5 of 7 donors.

Thus, stimulation of healthy donor PBMC with LAA mix generates CTL without depending on known HLA-restrictions and may be a practical approach for adoptive immunotherapy of myeloid leukemia after stem cell transplantation. Moreover, the strategy enabled the identification of several new class I and II-restricted peptides. Importantly, in vitro cytolytic activity could also be shown for MHC class II-restricted peptides, indicating the importance of CD4+ T cells for anti-leukemia responses.