P1001: modulation of mutant calreticulin-driven oncogenesis in myeloproliferative neoplasms by classical hla genes

Topic: 15. Myeloproliferative neoplasms - Biology & Translational Research Background: We have previously demonstrated that Human leukocyte antigen class I (HLA-I) and class -II (HLA-II) alleles can modulate JAK2 V617F-driven oncogenesis in myeloproliferative neoplasms (MPNs). Similar observations have been reported recently for CALR mutation (CALRmut)-positive MPNs. Aims: Here we aimed to further elucidate the putative role of HLA-I and –II genes CALRmut+ MPNs. Methods: We performed high-resolution genotyping of HLA-I and –II genes in a total of 42 CALRmut+ MPN patients, 158 JAK2 V617F+ MPN patients and 1083 healthy controls. We performed molecular dynamic simulations (MDS) of HLA-I: peptide binding and functional ELISPOT assays for immunogenicity of CALRmut peptides. Multivariate generalized linear models with age and gender as covariates were fitted for estimation of the association of each allele with the presence of CALR mutation versus healthy controls and JAK2 V617F+ MPNs using midasHLA package for Bioconductor. Associations with specific haplotypes were analyzed using the haplo.stat package for R. HLA-I alleles were tested for their theoretical ability to present CALRmut-derived peptides using the NetMHCPan4.1 server. Grantham distance between HLA alleles was performed using published Perl scripts. Molecular dynamics simulation studies were performed using GROMACs algorithm. ELISPOT assays for IFN-gamma release were used to assess the immunogenicity of two CALRmut-derived peptides. Results: To access the global effect of HLA class I genotype on the presence of JAK2 V617F mutation we calculated the patient harmonic-mean best rank (PHBR) for each subject as the harmonic mean of the best rank of NetMHCpan 4.1 predicted binding rank of all six HLA alelles. We did not find any difference between the PHBR of CALRmut+MPN patients and controls. We also estimated the Grantham distance for each HLA locus in heterozygote subjects, which is considered a measure for HLA alleles’ divergence. There was no correlation of the Grantham distance with disease status for HLA-I loci but the HED for HLA-DQB1 locus was significantly higher in CALRmut+MPNs. One HLA-I allele (HLA-C*06:02, p=0.046) was found inversely correlated with the presence of CALR mutation when age and gender of the subjects were also included in a generalized linear model. On the other hand a number of HLA-II alleles were found associated with CALRmut+, as follows: DQA1*01:02 (p = 0.003), DQB1*05:02 (p = 0.018), DQA1*04:01 (p = 0.02), DRB1*16:01 (p = 0.037), and DRB1*08:01 (p=-0.042). ELISPOT assays using stimulation with selected peptides of PBMC from healthy controls and MPN patients showed significantly stronger T-cell responses in controls with variable HLA-I genotype than CARLRmut or JAK2 V617F+ MPN patients. MDS analyses did not show conclusive results regarding the presentation of the peptides RRMMRTKMRM, RMMRTKMRM and SPARPRTSC by various HLA-I molecules. Therefore, in vitro binding assays are currently under way. Summary/Conclusion: This study provides the additional immunogenetic evidence that specific HLA class I and II molecules modulate CALRmut-driven oncogenesis. These findings suggest the necessity for further investigations to reveal the exact immune-modulatory mechanisms of HLA in MPNs, which may have important practical implications to the development of neoantigens-based vaccines in these disorders. Acknowledgements: This work was partially supported by NSF project KP-06-H41/2. Keywords: Myeloproliferative disorder, Immunogenicity, HLA