Hybrid Insulin peptides (HIPs) are neoepitopes formed by the fusion of two beta-cell granule proteins. HIPs are recognized by pathogenic CD4+ T-cells in the NOD mouse and by human islet-infiltrating CD4+ T cells. However, it is very technically challenging to identify new HIPs. Currently less than 20 HIPs recognized by human CD4+ T cells have been reported.
We set out to identify new HIPs formed by the fusion of fragments of proinsulin that are recognized by pancreatic islet-infiltrating CD4+ T cells isolated from deceased organ donors who had type 1 diabetes.
We generated, in silico, a library of the 9,409 HIPs that could theoretically be formed by fusion of any two 12mer fragments of human proinsulin. Then we filtered these candidate HIPs to select the top 20% of predicted HLA-DQ8 binders. Ultimately, we made a library of 4,488 12mers candidate HIPs. These were cloned into a bacterial expression vector and transformed into E. coli. Pools of HIP-expressing E. coli were screened for their capacity to stimulate Jurkat lines that expressed TCRs derived from human islet-infiltrating CD4+ T cells. A total of 110 TCR-expressing Jurkat lines were tested, from four T1D organ donors, against the library of 4,488 candidate HIPs. E. coli pools were screened until a single candidate HIP could be identified. Responses to candidate HIPs were then confirmed with synthetic peptides.
We identified 20 unique candidate HIPs recognized by 9 different TCRs, from two organ donors. For many TCRs we identified families of candidate HIPs, with partially conserved sequences, that stimulate the same clone.
This is the first functional, islet-infiltrating T-cell based, unbiased screening, to identify proinsulin derived HIPs. We demonstrate the feasibility of bacterial expression of T-cell antigens libraries and reveal a high degree of cross reactivity, for some TCRs, between new candidate HIPs.