UNC-GRK4-1: An Allele Specific Cancer Testis Antigen Identified Through Genomic Screening

Minor histocompatibility antigens (mHA) mediate immune responses that can cause both graft vs. host disease (GvHD) and the graft vs. leukemia (GvL) effect observed in allogeneic stem cell transplantation (SCT). These responses result from donor T-cells being exposed to non-self peptides that are presented by HLA molecules on recipient cells. In contrast, cancer testis antigens (CTA) are peptide antigens derived from proteins, normally expressed only in male germ cells, but also aberrantly expressed in some tumors. These antigens can induce T-cell responses in an autologous setting without using T-cells from an allogeneic donor. Because, mHA that are expressed by hematopoietic tissue or leukemia could be therapeutic targets, which would enhance GVL without increasing the risk of GVHD we performed a genomics based mHA screen to identify high-value mHA. However, one of our predicted antigens, UNC-GRK4-1, was only expressed in testis and human AML, and has properties more consistent with a CTA.

We used the Illumina NS-12 microarray platform to genotype 99 HLA-A0201 expressing myeloid leukemia patients and their HLA-matched donors at 13,917 non-synonymous coding single nucleotide polymorphisms (cSNPs). For both alleles in each of the 13,917 cSNPs we identified genetically predicted donor-recipient mHA responses, which were defined as the donor being homozygous for 1 allele and the recipient being either heterozygous or homozygous for the alternate allele. Using Fisher's exact test for each allele, we tested for the association between genetically predicted mHA responses and the clinical outcomes of remission vs. relapse over all 99 patients. Using the BioGPS database, we determined the tissue expression for the 261 alleles associated with remission to a p-value of <0.05. Five of these alleles were from genes expressed only in testis. Using Ensembl, we obtained the amino acid sequence that flanked each of these alleles, by 10 amino acids in both the N and C-terminal directions. We then used the IEDB ANN algorithm to predict candidate HLA-A0201 restricted epitopes for these peptides.

One of these alleles, rs1801058_T, located in the GRK4 gene, encoded a peptide (VLDIEQFSV, now called UNC-GRK4-1) that was predicted to bind tightly to HLA-A0201 (IC₅₀=18.7 nM), while the alternate allele, rs1801058_C (VLDIEQFSA) bound HLA-A0201 with less affinity (IC₅₀=125.21 nM). Using RT-PCR we confirmed GRK4 mRNA expression in human testis, and in 3 of 3 human AML samples. There was no GRK4 mRNA expression in human skin, liver or colon. We subsequently performed western blots and identified GRK4 protein in testis and 3 of 4 AML samples.

We used UNC-GRK4-1/HLA-A0201 tetramers to test for T-cell responses in post-SCT patients. We observed UNC-GRK4-1/HLA-A0201 tetramer-specific populations in 3 of 4 post-SCT samples evaluated (range 0.08% to 0.19% of CD8+ cells). Surprisingly, tetramer responses were observed in post-SCT samples where the donor was not predicted to produce a mHA response because his/her genotype contained at least 1 copy of the rs1801058_T allele indicating that central tolerance to UNC-GRK4-1 had not occurred in the donor T-cells prior to transplant.

In summary we have used a combined genomics and bioinformatics method to predict and confirm T-cell responses to a new leukemia antigen, UNC-GKR4-1. This antigen's properties are different from those of a mHA, and it is better classified as an allele-specific CTA where immune responses depend only on the genotype of the patient to produce the UNC-GRK4-1 peptide, not the donor. This distinguishes it from mHAs whose responses result from donor T-cells being exposed to non-self antigens expressed by the recipient. This distinction is important due to the fact that the maximum frequency of a mHA response in a SCT is only 25% because the immune response depends upon both donor and recipient genotypes. In contrast, because T-cell responses to UNC-GRK4-1 do not depend upon donor-recipient genotype mismatches, we predict that UNC-GRK4-1 specific responses could be elicited in any patients who express at least 1 copy of the rs1801058_T allele, whose allele frequency is 0.362. Hardy-Weinberg equilibrium predicts that 59% of patients will carry at least 1 copy of rs1801058_T and therefore express UNC-GRK4-1. Because of this high frequency of antigen expression and the leukemia-testis tissue restriction, UNC-GRK4-1 is a promising target for immunotherapy approaches.