Mapping of HLA class I binding motifs in forty-four fusion proteins involved in human cancers.

Chromosomal translocations coding for abnormal proteins are present in several human cancers. The junctional region of fusion proteins represents a potential target for a T cell-mediated antitumor response. T lymphocytes recognize antigens in the form of short peptides that must bind to HLA molecules. Different HLA specificities can bind different peptides, thus depicting different "peptide binding motifs." It would be useful to know whether a certain fusion protein presents, within its fusion region, the binding motif(s) for a certain HLA molecule. This information would allow a more focused immunological analysis. Here we present data obtained from the screening of the fusion regions of 44 different fusion proteins for the presence of binding motifs to 34 class I HLA molecules, including all of the most frequently encountered specificities. A total of 201 independent peptides was identified (range, 0-11 peptides/fusion protein). A marked heterogeneity among the 44 different fusion proteins analyzed is evident. For example, the pml/RARalpha fusion protein present in acute promyelocytic leukemia presents no binding motif (BCR 3) at all or to a single HLA molecule (Cw*0301, BCR 1). Alternatively, the fusion proteins BCR/ABL, ALL1/AF-6, EWS/ATF-1, or NPM/ALK exhibit motifs for several common HLA specificities. Heterogeneity is also present inside a single translocation (in ALL1/ENL, for example, different subtypes match motifs with cumulative frequencies in the population from 108 to 0%). In two cases where the relative frequency of different fusion protein subtypes was available, a tendency toward an inverse relationship between frequency and the percentage of population covered by the identified binding motifs was observed. Peptides with motifs for HLA A*0201, A3, and Cw*0702 were also tested for actual binding using a stabilization assay; 13-40% showed significant HLA binding, using this assay. However, fewer fusion protein-derived peptides bound to HLA A*0201 and A3 than non-fusion protein-derived peptides. These data provide the first list of peptides derived from fusion proteins that may be assessed as potential tumor-specific antigens.