Long term survival from AML reflects in large parts the ability to completely eradicate all AML cells. In most patients an initial remission can be achieved, but a small number of drug resistant cells often survive and mediate relapse. Both genetic (i.e. subclonal mutational heterogeneity) and epigenetic mechanisms likely play a role in the emergence of resistance.

In order to interrogate transcriptional states within bulk AML patient samples that might mediate epigenetic resistance, we conduced single cell RNA-Seq on 10 pediatric AML samples at initial diagnosis. Results were confirmed by analyzing publicly available data from 12 adult AML samples. We found surprisingly large heterogeneity in GATA2 expression both intra- and inter-patient. GATA2 is a transcription factor widely expressed throughout the hematopoietic system, with predominant expression in hematopoietic stem cells (HSCs) and myeloid progenitors. The role of GATA2 in pathogenesis of AML is complex, with overexpression, as well as gain, altered and loss of function mutations of GATA2 reported in AML. Several independent studies reported that high GATA2 expression is associated with worse outcomes. Intra-patient, GATA2 high expressing cells (GATA2high) skewed more immature, although both GATA2high and GATA2low cells were found at nearly all stages of maturation within the AML. When comparing both GATA2high and GATA2low cells within the same leukemia, pathways related to ribosomal biogenesis (RiBi) and protein translation were the most enriched in GATA2high subclones.

For functional studies, we turned to a Gata2 conditional retroviral KMT2A-MLLT3 mouse model. Similar to the patient samples, this model also showed both variable expression of Gata2 among different leukemic mice as well as within leukemia cells from the same mouse. Both high Gata2 (Gata2high) and low Gata2 (Gata2low) expressing clones were able to engraft and cause leukemia in recipient mice, although Gata2low clones did so with slightly lower penetrance and slightly longer latency. Similarly, deletion of Gata2 in bulk leukemia resulted in lower penetrance and longer latency, however, we found multiple animals with fully deleted, transplantable Gata2-/-AML showing that Gata2 is not absolutely required for leukemogenesis. Deletion of Gata2 from Gata2high murine AML cells resulted in significant downregulation of RiBi and protein synthesis pathways. Cut&Run identified binding of Gata2 at promoters of genes encoding ribosome proteins. This suggests a direct regulation of RiBi and protein homeostasis by Gata2. Multiple recent reports have implicated ribosomal proteins in the regulation of p53 mediated apoptosis. Indeed, we found that Gata2high murine AML cells showed a blunted p53 stabilization in response to the Mdm2 inhibitor nutlin-3, and Gata2high were more resistant to doxorubicin than Gata2low AML cells. Furthermore, in vitro treatment of bulk leukemias with Doxorubicin resulted in selection for Gata2high cells. Deletion of Gata2 sensitized leukemia cells to nutlin-3 and chemotherapy.

We next treated murine AML cells with the Pol1 inhibitor CX-5461, which inhibits RiBi. While Gata2high AML cells had been more resistant to doxorubicin, they were more sensitive to CX-5461 than Gata2low AML cells. In addition, we observed profound synergy between CX-5461 and doxorubicin.

In conclusion, we identified Gata2high subclones in patients with AML that likely represent a reservoir for resistance and relapse. Our functional data document high RiBi and a blunted p53 response in Gata2high AML cells that is dependent on Gata2. This cell state can be successfully targeted with a combination of standard chemotherapy and a Pol I inhibitor, nominating this combination as a promising therapeutic avenue for patients with AML.

Neff:Merck company: Current Employment. Bernt:Syndax: Research Funding; Epizyme: Patents & Royalties; Merck: Other: Husband is an employee of Merck and has stock.

Author notes


Asterisk with author names denotes non-ASH members.

Sign in via your Institution