Mechanisms of the CBFA2T3-GLIS2 Fusion in AML Maintenance

CBFA2T3-GLIS2 positive pediatric acute myeloid leukemia (AML) is among the poorest outcome subsets of childhood leukemia with overall survival barely reaching 10%. Mechanistically, as an aberrant transcription factor, the CBFA2T3-GLIS2 fusion directly alters transcription and impacts the activity of other key transcription factors leading to a block in differentiation and increased self-renewal. While the role of the fusion in disease initiation has been studied, its role in leukemia maintenance has been less well explored. With the development of new degradation approaches to "drug” pharmacologically challenging targets, the validation of CBFA2T3-GLIS2 as a candidate therapeutic target becomes critical. Recent genome-wide CRISPR screening has identified the genes involved in the fusion as strong, selective dependencies in the CBFA2T3-GLIS2 cell lines screened. We aimed to study the impact of deleting or degrading CBFA2T3-GLIS2 on the phenotypic and transcriptional state of AML.

We used a doxycycline-inducible knockout (Doxi-KO) system to deplete the fusion in three Cas9-transduced CBFA2T3-GLIS2 AML cell lines and one patient-derived-xenograft (PDX) model by targeting the fusion with two single guide RNAs (sgETO2 and sgGLIS2) and a non-targeting sgRNA (sgNT) as a negative control. We also transduced two non-CBFA2T3-GLIS2 AML cell lines with a sgRNA targeting ETO2 to exclude effects potentially induced by ETO2 wild-type depletion. Validation of the KO was assessed by western blot and TIDE-seq PCR. Using competitive-growth assays, we validated strong dependency on the fusion in the fusion-positive models. Accordingly, the three cell lines and the PDX harboring the fusion showed markedly impaired cell growth as assessed by ATP-based assays and colony forming ability in methylcellulose after Doxi-KO, whereas ETO2-WT KO alone did not lead to an impaired viability in the fusion-negative models. Moreover, after fusion knockout, transduced cells displayed monocytic (CD11b), megakaryocytic (CD41, CD42), or erythroid (CD71, CD235ab) differentiation markers by flow cytometry depending on the cell-of-origin and morphological changes consistent with differentiation.

To confirm our findings in vivo, we intravenously injected NSGS (NOD scid gamma SGM3) mice with CBFA2T3-GLIS2 positive WSU-AML cells either transduced with sgNT, sgETO2 or sgGLIS2. After confirmation of disease engraftment by bone marrow evaluation, doxycycline chow was initiated at day 14 to induce knockout of the fusion. At day 24, leukemia burden, reflected by the percentage of human CD45 cells, was significantly lower in sgETO2 and sgGLIS2 versus sgNT in peripheral blood, spleen, liver and bone marrow. Accordingly, survival was significantly prolonged in the sgETO2 and sgGLIS2 mice versus sgNT. Consistent with our in vitro observations, analysis of CD11b in human CD45+ cells revealed an increase in CD11b in the sgETO2 and sgGLIS2 groups compared to the sgNT group.

RNA sequencing of CBFA2T3-GLIS2 positive M07e AML cells after 5 days of Doxi-KO in vitro demonstrated enriched signal for myeloid and megakaryocytic differentiation. Specifically, we observed an enrichment in expression of the transcription factors GATA1 and GATA2 and a decrease in expression of GATA3. We next asked whether the increased expression of GATA1 was required for the differentiation observed with fusion knockout. Double knockout of GATA1 and CBFA2T3-GLIS2 was performed in two CBFA2T3-GLIS2 cell lines from different lineages. We observed a block in the induction of myeloid and megakaryocytic differentiation induced by fusion KO when GATA1 was concurrently knocked out, consistent with the hypothesis that repressed expression of GATA1 in CBFA2T3-GLIS2 AML is critical for the differentiation arrest observed in the disease.

In summary, we have validated a persistent dependency on CBFA2T3-GLIS2 in AML for disease maintenance in vitro and in vivo mediated in part by a block in differentiation through repressed GATA1, validating CBFA2T3-GLIS2 as a therapeutic target in this deadly disease.

Stegmaier:KronosBio: Consultancy, Research Funding; AstraZeneca: Consultancy; Auron Therapeutics: Consultancy, Current holder of stock options in a privately-held company, Membership on an entity's Board of Directors or advisory committees; Novartis: Research Funding.