Combination Targeted Therapy to Disrupt Aberrant Oncogenic Signaling and Reverse Epigenetic Dysfunction in IDH2- and TET2-Mutant Acute Myeloid Leukemia

Abstract

Genomic studies in acute myeloid leukemias (AML) have identified mutations that drive altered DNA methylation, including TET2 and IDH2 Here, we show that models of AML resulting from TET2 or IDH2 mutations combined with FLT3ITD mutations are sensitive to 5-azacytidine or to the IDH2 inhibitor AG-221, respectively. 5-azacytidine and AG-221 treatment induced an attenuation of aberrant DNA methylation and transcriptional output and resulted in a reduction in leukemic blasts consistent with antileukemic activity. These therapeutic benefits were associated with restoration of leukemic cell differentiation, and the normalization of hematopoiesis was derived from mutant cells. By contrast, combining AG-221 or 5-azacytidine with FLT3 inhibition resulted in a reduction in mutant allele burden, progressive recovery of normal hematopoiesis from non-mutant stem-progenitor cells, and reversal of dysregulated DNA methylation and transcriptional output. Together, our studies suggest combined targeting of signaling and epigenetic pathways can increase therapeutic response in AML.Significance: AMLs with mutations in TET2 or IDH2 are sensitive to epigenetic therapy through inhibition of DNA methyltransferase activity by 5-azacytidine or inhibition of mutant IDH2 through AG-221. These inhibitors induce a differentiation response and can be used to inform mechanism-based combination therapy. Cancer Discov; 7(5); 494-505. ©2017 AACR.See related commentary by Thomas and Majeti, p. 459See related article by Yen et al., p. 478This article is highlighted in the In This Issue feature, p. 443.

Conflict of interest statement

Conflicts of Interest: Jeremy Travis, Kim Straley, Camelia Gliser, Katharine Yen, and Craig B Thompson hold financial interests in Agios Pharmaceuticals.

©2017 American Association for Cancer Research.

Figures

Figure 1. Response of Tet2−/− Flt3ITD leukemia model to 5-Aza therapy

Tet2−/−Flt3ITD leukemia bone marrow transplant model mice were treated with vehicle (n=5) or 5-Aza (n=5) therapy. A, Histology of treatment response in liver, bone marrow, and spleen, infiltrating leukemic cells (red arrowhead), megakaryocytes (green arrowheads). B–F, Response to treatment in (B) WBC and spleen weight, (C) peripheral blood (PB) smears, (D) platelet, hematocrit, and PB Mac1−cKit+ population, (E) bone marrow stem cell differentiation with percentage of multipotent progenitors (MPP, CD48+CD150− gated on LSK, lin−Sca+cKit+), and (F) bone marrow erythroid differentiation with percentage of CD71+Ter119+ erythroid progenitors, WT- representative wild-type sample. Graphs and numbers indicate mean ± sem. G, WBC of mice following treatment with cycles of 5-Aza (n=5), holding of therapy (arrow), and resumption of 5-Aza. H, Graph of ERRBS analysis comparing vehicle (n=4) and 5-Aza treatment (n=4), indicating proportion of hypo-, unchanged, and hyper-methylated differentiated methylated cytosines (DMCs). I, Graph of overall genomic methylation proportion comparing WT LSKs to Tet2−/−Flt3ITD vehicle (green) and 5-Aza (red) treated LSKs. **p<.01, *p<.05 by t-test.

Figure 2. Idh2R140QFlt3ITD mouse model of leukemia

A, Targeting of endogenous Idh2 locus with Idh2R140Q mutation and the lox-Stop-lox (LSL) cassette. Genotyping PCR demonstrates Idh2WT, LSL Idh2R140Q, and excised alleles. B, Idh2R140Q Flt3ITD peripheral blood (PB) histology with blast cells (arrowhead), WBC (n=6 to 9) at 4–5 months, and spleen weight (n=4 to 6). Graph of mean ± sem. C, Immunophenotyping of PB Mac1 cKit (n=9 WT and Idh2R140QFlt3ITD), and bone marrow Myeloid progenitors (gated on lin−Sca1−cKit+) and stem cells (gated on lin−Sca1+cKit+) populations (n=3 WT, n=5 Idh2R140QFlt3ITD). Numbers indicate mean ± sem cKit+, Mac1+, GMP (CD34+CD16/32+), MEP (CD34−CD16/32−), and MPP (CD150−CD48+) percentages. Statistical comparisons to WT. D, GSEA analysis of differential expressed genes from RNA-seq of Idh2R140QFlt3ITD LSK (n=3) compared to WT LSK (n=3) against a human IDH2-mutant AML hypermethylated gene signature. **p<.01, *p<.05 by t-test.

Figure 3. Response of Idh2R140Q Flt3ITD leukemia model to AG-221 therapy

A, Idh2R140Q Flt3ITD transplant leukemia model following in vivo treatment trial with vehicle (n=7) or AG-221 (n=6) and serum 2-hydroxyglutarate (2HG) levels, WBC and peripheral blood (PB) cKit+ frequencies. Graphs of mean ± sem B, Representative histology of liver, spleen, and bone marrow after AG-221 treatment. (arrowhead indicates infiltrating leukemic cells). C, Graph of ERRBS analysis comparing vehicle (n=4) and AG-221 treatment (n=4), indicating proportion of hypo-, unchanged, and hyper-methylated differentiated methylated cytosines (DMCs). D, Graph of overall genomic methylation proportion comparing WT stem-progenitor cells lineage− Sca+ cKit+ (LSKs) to Idh2R140Q Flt3ITD vehicle (green line) and AG-221 (red line) treated LSKs. **p<.01, *p<.05 by t-test.

Figure 4. Differentiation response to therapy to 5-Aza and AG-221

A, Bone marrow (BM) lineage negative fraction in vehicle (n=5) or 5-Aza (n=5) treated Tet2−/−Flt3ITD leukemic model and vehicle or AG-221 (n=6) treated Idh2R140Q Flt3ITD leukemic model. Graphs of mean ± sd B, Leukemic fraction (CD45.2+) gated, immunophenotypic Mac1 Gr1 expression analysis of bone marrow cells from Tet2−/−Flt3ITD and Idh2R140Q Flt3ITD leukemia models in response to treatment and from WT mice (n=6). Numbers of Mac1+Gr1− percentage mean ± sem. Statistical comparisons between vehicle and treated group. C, Heat map of RNA-seq analysis of LSK CD45.2+ leukemia cells from Tet2−/−Flt3ITD (TF) mice treated with vehicle (n=3) or 5-Aza (n=4) and Idh2R140Q Flt3ITD (IF) mice treated with vehicle (n=3) or AG-221 (n=3) therapy, and also of wild-type (WT) (n=4) LSK cells. And graph of principle component analysis (PCA) of RNA seq with clustering of samples. D, GSEA analysis of RNA differential expression between vehicle and 5-Aza treated, and between vehicle and AG-221 treated LSK cells compared to a human IDH2-mutant AML methylated genes signature. E, BM CD45.2+ (leukemic) and CD45.1+ (WT) flow cytometry gating and quantitation from Tet2−/−Flt3ITD and of Idh2R140Q Flt3ITD leukemia models in response to treatment with 5-Aza and AG-221, respectively. Graphs of mean ± sem. **p<.01, *p<.05 by t-test.

Figure 5. Combination signaling and epigenetic therapy increases efficacy

A, Tet2−/−Flt3ITD leukemia treated with Vehicle, AC220, 5-Aza, or combination (n= 5 per group) with resulting WBC and spleen weight. B–C, Idh2R140Q Flt3ITD leukemia treated with Vehicle, AC220, AG-221, or combination (n=5 to 6 per group) with resulting (B) 2HG levels, WBC, spleen weight, and (C) histology of liver, spleen, and bone marrow (arrowhead indicate leukemic infiltration, green oval – splenic follicle). D–E, Differentiation of stem cell compartment (gated on LSK lin−Sca1+cKit+) from (D) Idh2R140Q Flt3ITD and (E)Tet2−/−Flt3ITD treated leukemia models. Graphs of mean ± sem. **p<.01, *p<.05 by t-test.

Figure 6. Enhanced methylation and leukemic response to combination signaling and epigenetic therapy

A, Graph of ERRBS analysis demonstrating differential hyper- and hypo- methylated cytosines (DMC) following treatment compared to vehicle in Tet2−/−Flt3ITD and Idh2R140Q Flt3ITD leukemia models (n=3). B, Heat map of genes demonstrating cooperative effects of AC220 and AG-221 therapy on methylation in the coding sequence in Idh2R140Q Flt3ITD leukemia model. C, Gata2 locus methylation status following treatment in Idh2R140Q Flt3ITD leukemia model. CpG indexed relative to Gata2 locus. D, Bone marrow CD45.1+ (normal) fraction following treatment in Tet2−/−Flt3ITD (n=5) and Idh2R140Q Flt3ITD (n=5–6) leukemia model. Graphs of mean ± sem. **p<.01, *p<.05 by t-test.