Reduced CBFB Expression Causes Compensatory Upregulation of RUNX1 Expression, Defining a Feedback Loop That May be Relevant for AML Pathogenesis

Core Binding Factor (CBF) is a heterodimeric transcription factor composed of one alpha subunit (predominantly RUNX1 in hematopoiesis), and one beta subunit (CBFB). RUNX1 is the DNA-binding subunit of the complex, while CBFB stabilizes the RUNX1-DNA interaction and protects the complex from ubiquitination and degradation. CBF is critical for hematopoiesis, and mice deficient in either CBFB or RUNX1 are known to have severe hematopoietic defects, including expanded hematopoietic progenitors, lymphopenia, and defective myeloid maturation, consistent with a preleukemic state. We recently showed that decreased CBFB expression (compared to healthy CD34 cells) is essentially universal in human AML samples, suggesting that decreased CBFB activity may be a broadly relevant mechanism for AML pathogenesis (Day et al JCI 2023). We also showed that RUNX1 mRNA is increased in virtually all AML samples, and that it is most significantly upregulated in cases with CBF pathway mutations, suggesting that this phenotype may represent a compensatory response to decreased CBF activity. To test this hypothesis directly, we performed bulk RNA-Seq on bone marrow cells from Cbfb^flox/flox x Vav1-Cre or wildtype mice (n=4 per group). As predicted, RUNX1 was indeed one of the most upregulated genes in Cbfb^flox/flox x Vav1-Cre bone marrow (2.9-fold increased, FDR < 0.01). Spectral flow cytometry showed multiple population shifts consistent with previously reported results, including decreased B cells, CD4+ T cells, dendritic cells, and immature erythroid cells, as well as increased myeloid progenitors and ST-HSCs. However, we also detected a previously-unreported, novel myeloid progenitor population (Lin- Kit+ Sca- CD34- CD16/32+) that is phenotypically similar to the preleukemic population described in a CBFB::MYH11 knock-in murine model (Zhen et al. Blood 2020). To confirm and extend these observations, we performed scRNA-Seq on bone marrow cells from 2 Cbfb^flox/flox x Vav1-Cre and 5 wildtype control mice. We noted striking changes in multiple transcriptionally-defined populations, including decreased B cells, CD8+ T cells, and dendritic cells, and increased mature neutrophils and MEPs, as well as a transcriptionally-distinct population of cells with hematopoietic progenitor features that is now being further characterized. However, even within transcriptionally-defined subpopulations, Cbfb^flox/flox x Vav1-Cre cells clustered independently from their wildtype counterparts, suggesting that they have distinct transcriptional programs. We observed increased RUNX1 expression in several CBFB deficient hematopoietic populations, including megakaryocytes, neutrophil precursors, MEPs, GMPs, MPPs, and ST-HSCs, indicating that the RUNX1 feedback loop is relevant in multiple hematopoietic subpopulations. To determine whether restoration of CBFB expression can drive RUNX1 expression down to normal levels, we used retroviral vectors to express CBFB in Cbfb^flox/flox x Vav1-Cre hematopoietic cells. Indeed, restoration resulted in complete normalization of RUNX1 mRNA within 7 days. Overall, these data support the hypothesis that elevated RUNX1 expression in Cbfb-deficient mouse hematopoietic cells—and likely also in human AML cells—reflects a compensatory response to reduced CBF activity, and strongly suggests that restoring CBFB expression may represent a broadly relevant therapeutic approach in human AML cells. Experiments to test the effects of CBFB “addback” in primary human AML cells are in progress.