Mutational Landscape of the Transcriptome Offers Putative Targets for Immunotherapy of Myeloproliferative Neoplasms

Fiorella Schischlik, Roland Jäger, Felix Rosebrock, Eva Hug, Michael K Schuster, Raimund Holly, Elisabeth Fuchs, Jelena D. Milosevic Feenstra, Edith Bogner, Bettina Gisslinger, Martin Schalling, Elisa Rumi, Daniela Pietra, Gottfried F Fischer, Ingrid Faé, Loan Vulliard, Jörg Menche, Torsten Haferlach, Manja Meggendorfer, Anna Stengel, Christoph Bock, Mario Cazzola, Heinz Gisslinger and Robert Kralovics

Key Points

  • Driver mutations in CALR or MPL encode for predicted neoantigens that bind MHC class I with high affinity in MPN patients.

  • A high number of MPN patients show evidence of recurrent candidate neoantigens, suggesting a potential use for targeted immunotherapy.


Ph-negative myeloproliferative neoplasms (MPNs) are hematological cancers subdivided into entities with distinct clinical features. Somatic mutations in JAK2, CALR, and MPL have been described as drivers of the disease, together with a variable landscape of non-driver mutations. Despite detailed knowledge of disease mechanisms, targeted therapies effective enough to eliminate MPN cells are still missing. In this study, we aimed to comprehensively characterize in 113 MPN patients the mutational landscape of the granulocyte transcriptome using RNA-seq data and subsequently examine the applicability of immunotherapeutic strategies for MPN patients. Following implementation of customized workflows and data filtering, we identified a total of 13 (12/13 novel) gene fusions, 231 non-synonymous SNVs, and 21 Indels in 106/113 patients. We found a high frequency of SF3B1 mutated PMF patients (14%) with distinct 3' splicing patterns, many of these with a protein altering potential. Finally, from all mutations detected, we generated a virtual peptide library and used NetMHC to predict 149 unique neoantigens in 62% of MPN patients. Peptides from CALR and MPL mutations provide a rich source of neoantigens due to their unique property to bind many common MHC class I molecules. Finally, we propose that mutations derived from splicing defects present in SF3B1 mutated patients may offer an unexplored neoantigen repertoire in MPN. We validated 35 predicted peptides to be strong MHC class I binders through direct binding of predicted peptides to MHC proteins in vitro. Our results may serve as a resource for personalized vaccine or adoptive cell-based therapy development.

  • Submitted October 29, 2018.
  • Revision received April 20, 2019.
  • Accepted April 19, 2019.