TCR-T cells expressing a TCR recognizing the MPO propeptide-derived MPO132-140:HLA-a*02:01 specifically kill AML and spare healthy myeloid cells Free

Patients with AML/MDS have limited therapeutic options. While antigen-specific T cell therapies (e.g. CAR-T) have shown curative potential in treating lymphoid leukemias, their use to treat myeloid leukemia has been challenging. This is despite the observation that AML/MDS can be susceptible to immune-mediated killing, as evidenced by the graft-vs-leukemia effect observed after allogeneic stem cell transplant. A major barrier to antigen-specific T cell therapy in AML/MDS is the identification of suitable leukemia target antigens. A target antigen must be highly expressed by malignant cells to drive efficacy but not expressed by normal myeloid cells to avoid myeloablation.

Myeloperoxidase (MPO) is an intracellular, lineage-defining marker of myeloid cells, and robust MPO protein expression is found in both normal and malignant myeloid cells. MPO participates in anti-microbial responses by producing toxic compounds such as hypochlorous acid and reactive oxygen species, and its function and intracellular localization are tightly controlled to avoid host tissue damage. Synthesis of MPO occurs in a well-defined, stepwise process; enzymatically inactive preMPO is produced in myeloblasts, and undergoes glycosylation, proteolytic cleavage, and packaging into azurophilic granules. While fully differentiated myeloid cells (e.g. neutrophils and monocytes) contain large amounts of mature MPO stored in granules, de novo MPO production does not occur in these cells. Conversely, persistent MPO production is observed in many leukemias, and an excess of MPO+ cells in bone marrow is diagnostic of AML.

We hypothesized that the N-terminal 165 amino acids of preMPO, which are removed during the MPO maturation process, may produce leukemia-restricted peptide:HLA antigens. Analysis of immunopeptidomics datasets detected HLA class I-bound MPO_132-140(SLWRRPFNV) in HLA-A*02:01+ patient AML samples, but not in HLA-A*02:01+ healthy tissues nor HLA-A*02:01- patient AML samples. UV-mediated HLA ligand exchange assays were performed and demonstrated that MPO_132-140robustly binds HLA-A*02:01. We assessed if MPO_132-140:HLA-A*02:01 could be effectively recognized by T cells. Functional T cell expansion assays were used to expand MPO_132-140:HLA-A*02:01-specific TCRs from non-HLA-A2+ donor CD8 T cell repertoires, and MPO_132-140:HLA-A*02:01-tetramer binding cells were sorted and TCR sequenced. TCR sequences were inserted into retroviral constructs and used to make TCR-T cells for functional testing.

TCR-T activation assays with peptide pulsed target cells identified a TCR with a sub-1nM EC₅₀, similar in potency to anti-viral TCR-T cells. Cytotoxicity experiments revealed that MPO_132-140:HLA-A*02:01-specific TCR-T cells robustly killed HLA-A2+MPO+ leukemia cell lines and primary patient-derived AML, demonstrating that MPO_132-140:HLA-A*02:01 is a naturally processed and presented antigen. No killing was observed when HLA-A2+MPO- or HLA-A2-MPO+ cells were used as targets, demonstrating the specificity of the TCR. MPO_132-140:HLA-A*02:01 TCR-T cells did not kill normal MPO+ HLA-A2+ neutrophils and monocytes (despite robust MPO protein expression) demonstrating that the MPO_132-140:HLA-A*02:01 is not expressed by mature myeloid cells at immunogenic levels. Colony formation assays demonstrated that MPO_132-140:HLA-A*02:01-specific TCR-T cells also spare normal myeloid progenitor cells, despite their ability to produce preMPO (containing the MPO_132-140 peptide). Targeting of malignant cells over normal progenitors may be explained by differences in MPO expression, glycosylation, and/or subcellular localization - as has been described for other azurophilic granule proteins (e.g. proteinase 3 and neutrophil elastase).

Collectively, this data shows that MPO_132-140:HLA-A*02:01 is a myeloid leukemia-restricted antigen, and that T cells targeting it can kill leukemia while sparing normal myeloid cells. Targeting of MPO_132-140:HLA-A*02:01 with CAR-T, TCR-like antibodies, and bi-specifics is a rational extension of these findings. Furthermore, while HLA-A*02:01 is the most common HLA-A allele in Americans, the propeptide of MPO contains predicted binders for other many other common HLA alleles. Given the results shown here and the biology of MPO maturation, other MPO propeptide derived peptide:HLA may be suitable targets for non-HLA-A2 leukemias as well.