Identification of Leukemia Associated Antigens From ANKRD17 and CDK4 Using Mass Spectrometry Based Screening,

The discovery of therapeutically relevant tumor antigens requires the demonstration of endogenous antigen processing and presentation by class I HLA molecules and evidence for the antigen's ability to elicit a specific immune response. Traditional antigen discovery methods have focused on initially generating T cell clones that recognize a target with subsequent testing of sequentially purified peptide pools to determine the chemical structure of the antigen. Generating T cell clones and the subsequent antigen discovery processes are labor-intensive, time-consuming and inefficient because antigen characteristics such as source protein, tissue restriction, HLA-restriction, and binding affinity cannot be determined prior to the generation of the T cell clone. To address these limitations, we developed an antigen discovery method based upon initial identification of HLA-epitopes with subsequent immune characterization that bypasses the need to generate a T cell clone and enabled us to identify and characterize 2 leukemia associated antigens within 1 year.

Wild-type HLA-A0201 was stably transduced into the U937 myeloid leukemia cell line. Approximately 2×10⁹ HLA-A0201+U937 were lysed and the peptide/HLA-A0201 complexes were purified by immunoaffinity column chromatography. The peptides were liberated from the HLA by weak acid elution and analyzed by reverse phase HPLC/tandem mass spectrometry (MS/MS). Collision induced dissociation (CID) MS/MS data were analyzed using the Mascot search engine. From 23 unambiguously sequenced, high-abundance peptides we analyzed 2 peptides, derived from ANKRD17 (UNC-A17 = LLIERGASL) and CDK4 (UNC-CK4 = ALTPVVVTL). These peptides were selected because ANKRD17 is a known source protein for the breast cancer antigen NY-BR-16, and CDK4 initiates entry into S-phase and is currently an investigational therapeutic target for treatment of acute myeloid leukemia (AML).

We confirmed CDK4 protein expression and observed variable (1 sample with high expression and 2 samples with low expression) expression of ANKRD17 in 3 of 3 human AML samples using western blots. Using an in-vitro assay, we found that both antigens bound HLA-A0201 with a dissociation t_1/2 > 8 hours. Because of this slow off rate, we performed T cell expansion studies from healthy donor peripheral blood mononuclear cells (PBMC). We were able to generate antigen-specific oligoclonal T cell populations for both antigens. These T cell populations demonstrated antigen-specific immune function based upon intracellular interferon-γ flow cytometry measurements using both peptide-pulsed autologous dendritic cells and HLA-A0201+U937 cells as immune targets. We next looked for evidence of in-vivo T cell immunity towards these antigens in 5 patients (2 AML, 1 CML, 1 MDS, 1 MF) who had undergone allogeneic stem cell transplant. Using UNC-A17 and UNC-CK4 peptide loaded, fluorescently conjugated HLA-A0201 tetramers we identified UNC-CK4 specific T cells in 2 samples (0.22% and 0.39% of total CD8+ cells) and UNC-A17 specific T cells in 1 sample (0.64% of total CD8+ cells).

Using this novel methodology we were able to identify 2 new leukemia associated antigens by directly examining the target cells for antigen expression and epitope processing. This direct methodology eliminated the need to generate a T cell clone or evaluate an entire peptide pool for evidence of immunogenicity, as the first step in antigen discovery. Refinement of our HPLC-MS techniques will enable us to sequence increasing numbers of peptide epitopes from the source cells. These technical improvements combined with the standard genomics and bioinformatics tools that are currently available will allow us to narrow our focus on the discovery of immunologically relevant tumor associated self antigens and minor histocompatibility antigens that have very narrow tissue expression and favorable HLA binding characteristics.

No relevant conflicts of interest to declare.