Our preliminary studies identified recurrent, somatic mutations in Kdm6a in a mouse model of acute promyelocytic leukemia (APL). Putative inactivating mutations of Kdm6a (primarily small deletions) occurred in approximately 50% of mouse APL samples, making these the most common acquired progression events in this model. We found somatic mutations of KDM6A in 4 of 200 acute myeloid leukemia (AML) samples. However, in a cohort of 42 patients with APL, we identified no somatic alterations in KDM6A . To better understand the role of Kdm6a in malignant hematopoiesis, we intercrossed our Kdm6a conditional knockout (KO) x Vav1 - Cre mice with a "knock-in" mouse model of APL (Ctsg-PML-RARA mice). To determine whether Kdm6a inactivation cooperates with PML - RARA, we performed a tumor watch that included all genotypes of Kdm6a conditional KO female and male triple cross mice. After 18 months of observation, we found that the latency of APL was significantly decreased in all three cohorts of Kdm6a conditional KO x Vav1 - Cre x Ctsg-PML-RARA mice (Figure 1). This finding was most pronounced in female homozygous Kdm6a KO triple cross mice, which had a median survival of 161 days versus 365 days in the control cohort (P < 0.01 by log-rank test). The Kdm6a -null mice developed the typical phenotype of mouse APL, including splenomegaly, myeloproliferation and leukemic cells that coexpressed Gr-1 and Kit. The triple cross APL cells remained sensitive to all- trans retinoic acid (ATRA) and arsenic trioxide ex vivo . Compared with 56% penetrance in Ctsg-PML-RARA mice, female homozygous and male hemizygous Kdm6a KO x Vav1 - Cre x Ctsg-PML-RARA mice developed APL with 100% penetrance. We fully characterized pre-leukemic hematopoiesis in young (6 to 8-weeks-old) Kdm6a conditional KO x Vav1 - Cre x Ctsg-PML-RARA mice (compared to appropriate littermate controls) to define the underlying mechanism(s) by which Kdm6a inactivation contributes to leukemogenesis. Like our double cross Kdm6a conditional KO x Vav1 - Cre mice, the loss of Kdm6a expression in young female homozygous triple cross mice caused mild thrombocytopenia and splenomegaly. These young mice had neutrophilia and lymphopenia, which the female Kdm6a -null double cross mice only acquired with age. We observed a trend for increased numbers of colonies in a myeloid serial replating assay that was also specific to this cohort. Next, we demonstrated that this cohort had significant decreases in the Kit+Lin-Sca-1+ (KLS) cell, short-term hematopoietic stem cell (ST-HSC), and multipotent progenitor (MPP) hematopoietic stem and precursor compartments. This recapitulated the findings from the young, double cross Kdm6a -null female mice but was more pronounced. We found a concomitant increase in granulocyte-macrophage progenitors (GMPs) and decrease in megakaryocyte-erythroid progenitors (MEPs) that was also specific for the female Kdm6a -null triple cross cohort. Finally, we performed gene expression profiling via exon arrays on flow-sorted SLAM cells from aged (1-year-old) female Kdm6a -null double cross mice and aged controls. Unsupervised hierarchical clustering revealed a clear distinction between cohorts. We identified 1520 differentially expressed genes by an ANOVA analysis (fold change > 2, P < 0.05) between the Kdm6a -null versus WT aged SLAM cells. The list of differentially expressed genes included Pml (downregulated in Kdm6a -null SLAM cells with fold change of 2.35) and Flt3 (upregulated in Kdm6a -null SLAM cells with fold change of 2.32). Expression changes of these two genes may be associated with the mechanism of how inactivation of Kdm6a cooperates with PML-RARA in our mouse model. Pml haploinsuffiency has been show to modestly cooperate with PML-RARA in mouse models of APL, and FLT3 mutations are the most common secondary mutations in APL. In an effort to further elucidate the mechanism by which loss of Kdm6a contributes to malignant hematopoiesis, we have completed RNA-Seq, ATAC-Seq, anti-histone H3 (monomethyl K4) ChIP-Seq, and anti-histone H3 (trimethyl K27) ChIP-Seq studies on mouse APL samples derived from Ctsg-PML-RARA mice that developed spontaneous leukemias (n = 3 Kdm6a -WT samples versus n = 3 Kdm6a -mutant samples). Analysis of this data set is now underway. In sum, our data have demonstrated a functional role of Kdm6a in malignant hematopoiesis and have improved our understanding of its subsequent connection to leukemogenesis.


No relevant conflicts of interest to declare.

Author notes


Asterisk with author names denotes non-ASH members.