Abstract
Introduction. Acute erythroid leukemia (AEL) is a high-risk leukemia subtype of poorly understood genetic basis. In integrated comparative genomic analysis of 159 AEL and 1509 non AEL myeloid tumors, we identified 5 age-related AEL subtypes with distinct genomic features and outcome: adult AEL with bi-allelic alterations in TP53 (31%), frequently co-occurring with alterations in DNMT3A,BCOR, EZH2, RB1, or NFIX; NPM1-mutated (12%); KMT2A-mutated/rearranged (11%), mostly co-mutated with STAG2; pediatric, NUP98-rearranged (5%) and adult, DDX41-mutated (3%). Thirty-eight percent of cases lacked an identifiable exclusive recurrent founding alteration but were enriched in mutations in ASXL1, TET2 and splicing factors.
Methods. To explore the roles and cooperativity of the identified alterations in leukemogenesis we used CRISPR-Cas9 genome editing to induce combinations of loss-of-function mutations in 9 recurrently mutated genes in AEL (Tp53, Tet2, Dnmt3a, Asxl1, Ezh2, Stag2, Bcor, Ppm1d, Rb1 and Nfix). Based on patterns of mutation association, we generated 6 pools of lentiviral vectors (Table 1) with different combinations of single guide RNA (sgRNA) to induce multiplex genome editing in Cas9-eGFP-mouse lineage-negative hematopoietic stem cells (HSCs). Transduced cells were transplanted into lethally irradiated congenic mice. Tumors were characterized by morphology, immunophenotyping, and genomic analysis (sequencing of sites of editing, and exome, methylation and transcriptome sequencing). Ex vivo drug screening was performed to test sensitivity to 192 therapeutic agents, including conventional chemotherapeutic agents, and compounds targeting epigenetic regulators, protein kinases and cell cycle checkpoints.
Results. In contrast to the uniform representation of guide RNAs observed in HSCs pre-transplant, tumors exhibited enrichment of specific sgRNAs with tumors of specific phenotype. We frequently observed bi-allelic editing of Tp53, Bcor, Tet2, Dnmt3a, Rb1 and Nfix in agreement with the presence of bi-allelic loss in patients. Concomitant editing and inactivation of Tp53/Bcor/Dnmt3a, or Tp53/Bcor/Rb1/Nfix promoted the development of acute erythroid leukemia (GATA1+/RUNX1+/GlyA+/-Ter119+/- and B220/CD19/PAX5/CD3/Mac1/Gr1-). Concomitant editing of Tp53/Bcor/Tet2 promoted the development of B-lineage ALL, and editing of Dnmt3a and Tet2 without Tp53 promoted T-cell ALL. Leukemic clones from primary tumors were serially transplantable across multiple different genetic backgrounds, with the same dominant clone present in all transplanted mice. Notably, mice that did not develop leukemia showed enrichment of different combinations of sgRNAs for Tet2, Dnmt3a and/or Asxl1, genes commonly mutated in clonal hematopoiesis of indeterminate potential, confirming that these mutations are alone not sufficient to induce leukemogenesis. Additional somatic mutations were acquired during clonal expansion of leukemic cells such as alterations of Notch1 and Ikzf1 in T-ALL; Setd2 at the serial passaging of T-ALL; Ptpn11, Kit (D816V), Kras (Q61L) and Lmo7 in the AEL models; and Sf3b3 in the B-ALL model (Fig 1A). Tumors with mutated Tp53 acquired aneuploidy whereas Tet2-mutated cells were genomically stable (Fig. 1B). Unsupervised hierarchical clustering of gene expression profiling identified 3 subgroups (Fig. 1C) associated with distinct genotypes and methylation profiles (Fig. 1D). Tet2-mutated tumors showed increase of hypermethylated sites and co-mutated Bcor/Dnmt3a leukemic cells showed loss of methylation. Eleven tumors representative of key AEL genotypes from the established models were selected to explore drug sensitivity. Response to individual drugs was associated with genotype with co-mutated Tet2/Dnmt3a T-ALL cells sensitive to bromodomain and histone methyltransferase inhibitors; co-mutated Bcor/Tp53/Dnmt3a or Bcor/Rb1 AEL cells to CDK9 inhibitor (LY2857785); Tp53 mutations alone were exclusively sensitive only to PARP inhibition (Talazoparib). In vivo mouse studies are ongoing to confirm ex vivo results.
Conclusions: We successfully generated genetically defined models of AEL, demonstrated the role of Tp53 and Bcor mutations in driving the erythroid phenotype, and showed that sensitivity to different classes of compounds is genotype-dependent. These results provide the rational for testing new targeted agents in AEL.
Disclosures
Mullighan: Abbvie: Research Funding; Amgen: Honoraria, Speakers Bureau; Loxo Oncology: Research Funding; Cancer Prevention and Research Institute of Texas: Consultancy; Pfizer: Honoraria, Research Funding, Speakers Bureau.
Clinical Genomic Analysis and Diagnosis --Genomic Analysis Ex Vivo, in Vitro and in Silico