Pipeline for Discovering Neoepitopes Generated By Alternative Splicing in B-ALL

Pediatric B-cell acute lymphoblastic leukemias (B-ALL) are striking for their low mutational burdens. Ostensibly, this paucity of mutation-associated neoantigens could limit the development of novel immunotherapies. However, our lab has previously shown that globally aberrant splicing is a hallmark of B-ALL, as compared to normal pro-B cells. Here we hypothesize that alternative splicing represents a new source of targetable neo-epitopes, distinct from missense mutations.

To test this hypothesis, we constructed a pipeline for neo-epitope discovery, validation, and therapeutic development. First, we obtained a very large RNA-seq dataset covering over 400 pediatric B-ALL samples from St. Jude Children's Research Hospital. For normal controls, RNA-seq was performed on 4 pro-B cell fractions from bone marrow donors. We then applied a bioinformatic splicing analysis, MAJIQ 2.0, to deeply mine the dataset for local splicing variants (LSVs) unique to B-ALL. To bypass MHC presentation barriers, we focused on LSVs mapping to exons encoding extracellular protein domains (ectoLSVs). This filtering yielded a list of 914 ectoLSVs in 430 genes. These ectoLSVs were further filtered for those that preserved opening reading frames, occurred with high prevalence, and demonstrated a large differential expression between B-ALL and normal controls.

One such prominent event was the increased skipping in B-ALL of CD22 of exon 5-6 (Δex5-6). We validated this event at the RNA level with RT-PCR in 18 out of 18 primary B-ALL samples from CHOP Biobank. Using Nanopore-based long read RNA sequencing, we confirmed the Δex5-6 event exists within a larger translatable transcript. BALB/c mice were then immunized against a peptide containing the junction site of exon 4 and 7 to generate hybridomas; several monoclonal antibody (mAb) against CD22 Δex5-6 were successfully obtained. To validate the specificity of our mAbs, we used CRISPR to delete endogenous CD22 in OCI-Ly10 cells and reconstituted the cells with either CD22 Δex5-6 or full length CD22. The mAb demonstrated remarkable specificity for CD22 Δex5-6 and did not bind to the full-length isoform. We also demonstrated that at least one of these mAbs (clone 3A3) specifically binds to endogenously expressed CD22 Δex5-6 in multiple B-ALL cell lines by immunoprecipitation and Western blot analysis. Our next steps are to develop 3A3-based antibody-drug conjugates and chimeric antigen receptors for further therapeutic testing.

In summary, our results validate that 1) our current discovery pipeline is able to identify targetable splicing-derived neo-epitopes, and that 2) antibodies with impressive specificity can be generated against such neo-epitopes. Furthermore, this new paradigm has the promise of increasing the repertoire of highly specific immunotherapy targets in B-ALL, despite its low mutation burden. Of note, this strategy could also be carried forward into therapeutic development for many other cancers beyond B-ALL.