Development of combination therapies with BTK inhibitors and dasatinib to treat CNS-infiltrating E2A-PBX1+/preBCR+ ALL.

ABSTRACT: The t(1;19) translocation, encoding the oncogenic fusion protein E2A (TCF3)-PBX1, is involved in acute lymphoblastic leukemia (ALL) and associated with a pre-B-cell receptor (preBCR+) phenotype. Relapse in patients with E2A-PBX1+ ALL frequently occurs in the central nervous system (CNS). Therefore, there is a medical need for the identification of CNS active regimens for the treatment of E2A-PBX1+/preBCR+ ALL. Using unbiased short hairpin RNA (shRNA) library screening approaches, we identified Bruton tyrosine kinase (BTK) as a key gene involved in both proliferation and dasatinib sensitivity of E2A-PBX1+/preBCR+ ALL. Depletion of BTK by shRNAs resulted in decreased proliferation of dasatinib-treated E2A-PBX1+/preBCR+ cells compared with control-transduced cells. Moreover, the combination of dasatinib with BTK inhibitors (BTKi; ibrutinib, acalabrutinib, or zanubrutinib) significantly decreased E2A-PBX1+/preBCR+ human and murine cell proliferation, reduced phospholipase C gamma 2 (PLCG2) and BTK phosphorylation and total protein levels and increased disease-free survival of mice in secondary transplantation assays, particularly reducing CNS-leukemic infiltration. Hence, dasatinib with ibrutinib reduced pPLCG2 and pBTK in primary ALL patient samples, including E2A-PBX1+ ALLs. In summary, genetic depletion and pharmacological inhibition of BTK increase dasatinib effects in human and mouse with E2A-PBX1+/preBCR+ ALL across most of performed assays, with the combination of dasatinib and BTKi proving effective in reducing CNS infiltration of E2A-PBX1+/preBCR+ ALL cells in vivo.

DNMT and HDAC inhibition induces immunogenic neoantigens from human endogenous retroviral element-derived transcripts.

Immunotherapies targeting cancer-specific neoantigens have revolutionized the treatment of cancer patients. Recent evidence suggests that epigenetic therapies synergize with immunotherapies, mediated by the de-repression of endogenous retroviral element (ERV)-encoded promoters, and the initiation of transcription. Here, we use deep RNA sequencing from cancer cell lines treated with DNA methyltransferase inhibitor (DNMTi) and/or Histone deacetylase inhibitor (HDACi), to assemble a de novo transcriptome and identify several thousand ERV-derived, treatment-induced novel polyadenylated transcripts (TINPATs). Using immunopeptidomics, we demonstrate the human leukocyte antigen (HLA) presentation of 45 spectra-validated treatment-induced neopeptides (t-neopeptides) arising from TINPATs. We illustrate the potential of the identified t-neopeptides to elicit a T-cell response to effectively target cancer cells. We further verify the presence of t-neopeptides in AML patient samples after in vivo treatment with the DNMT inhibitor Decitabine. Our findings highlight the potential of ERV-derived neoantigens in epigenetic and immune therapies.

In vivo kinetics of early, non-random methylome and transcriptome changes induced by DNA-hypomethylating treatment in primary AML blasts.

Despite routine use of DNA-hypomethylating agents (HMAs) in AML/MDS therapy, their mechanisms of action are not yet unraveled. Pleiotropic effects of HMAs include global methylome and transcriptome changes. We asked whether in blasts and T-cells from AML patients HMA-induced in vivo demethylation and remethylation occur randomly or non-randomly, and whether gene demethylation is associated with gene induction. Peripheral blood AML blasts from patients receiving decitabine (20 mg/m2 day 1-5) were serially isolated for methylome analyses (days 0, 8 and 15, n = 28) and methylome-plus-transcriptome analyses (days 0 and 8, n = 23), respectively. T-cells were isolated for methylome analyses (days 0 and 8; n = 16). We noted massive, non-random demethylation at day 8, which was variable between patients. In contrast, T-cells disclosed a thousand-fold lesser, random demethylation, indicating selectivity of the demethylation for the malignant blasts. The integrative analysis of DNA demethylation and transcript induction revealed 87 genes displaying a significant inverse correlation, e.g. the tumor suppressor gene IFI27, whose derepression was validated in two AML cell lines. These results support HMA-induced, non-random early in vivo demethylation events in AML blasts associated with gene induction. Larger patient cohorts are needed to determine whether a demethylation signature may be predictive for response to this treatment.

Integrative study of EZH2 mutational status, copy number, protein expression and H3K27 trimethylation in AML/MDS patients.

BACKGROUND: Mutations in the EZH2 gene are recurrently found in patients with myeloid neoplasms and are associated with a poor prognosis. We aimed to characterize genetic and epigenetic alterations of EZH2 in 58 patients (51 with acute myeloid leukemia and 7 with myelodysplastic or myeloproliferative neoplasms) by integrating data on EZH2 mutational status, co-occurring mutations, and EZH2 copy number status with EZH2 protein expression, histone H3K27 trimethylation, and EZH2 promoter methylation. RESULTS: EZH2 was mutated in 6/51 acute myeloid leukemia patients (12%) and 7/7 patients with other myeloid neoplasms. EZH2 mutations were not overrepresented in patients with chromosome 7q deletions or losses. In acute myeloid leukemia patients, EZH2 mutations frequently co-occurred with CEBPA (67%), ASXL1 (50%), TET2 and RAD21 mutations (33% each). In EZH2-mutated patients with myelodysplastic or myeloproliferative neoplasms, the most common co-mutations were in ASXL1 (100%), NRAS, RUNX1, and STAG2 (29% each). EZH2 mutations were associated with a significant decrease in EZH2 expression (p = 0.0002), which was similar in patients with chromosome 7 aberrations and patients with intact chromosome 7. An association between EZH2 protein expression and H3K27 trimethylation was observed in EZH2-unmutated patients (R2 = 0.2, p = 0.01). The monoallelic state of EZH2 was not associated with EZH2 promoter hypermethylation. In multivariable analyses, EZH2 mutations were associated with a trend towards an increased risk of death (hazard ratio 2.51 [95% confidence interval 0.87-7.25], p = 0.09); similarly, low EZH2 expression was associated with elevated risk (hazard ratio 2.54 [95% confidence interval 1.07-6.04], p = 0.04). CONCLUSIONS: Perturbations of EZH2 activity in AML/MDS occur on different, genetic and non-genetic levels. Both low EZH2 protein expression and, by trend, EZH2 gene mutations predicted inferior overall survival of AML patients receiving standard chemotherapy.