Transcriptional plasticity is an evolving phenomenon in cancer biology. Mutational profiling alone may not suffice to dissect transcriptional dependency and underlying epigenetic vulnerabilities in tumorigenesis. Histone 3 lysine 27 (H3K27) demethylases (UTX, UTY and JMJD3) critically regulate transcriptional architecture. Recently it has been demonstrated that Utx (Kdm6a) plays tumor suppressor role in myeloid leukemogenesis through noncatalytic activity (Gozdecka M, et al., Nat Genet, 2018). Conditional (Mx1-Cre) deletion of Utx caused development of acute myeloid leukemia (AML) in ~ 60% of the mice; wherein, only Utx-/-, but not Utx+/-, aged (22 months) mice presented with AML. Paradoxically, previous report suggested that Utx conditional (Vav1-Cre) knock out male/female mice did not develop leukemia over 18 months (Tian, L., et al., Blood, 2015). In our recent report we have identified that expression of UTX is significantly increased in human primary AML, and pharmacological inhibition of H3K27 demethylase catalytic activity attenuated survival of AML cells (Boila L.D. et al,. Exp Hematol, 2017). Therefore the contribution of UTX in AML pathogenesis remains context dependent, and probably contentious, and warrants further investigations.
ATP-dependent chromatin remodelers have been implicated in AML pathobiology (Chatterjee S.S., et al., Mol Cancer Res, 2018). We reported that loss of MBD3, a scaffold of NuRD chromatin remodeler, in human primary AML cells associates with nucleation of leukemic NuRD (Biswas M et al., Blood, 2017). Loss of Mbd3(Vav1-Cre) has been shown to disrupt NuRD complex integrity and causes T-cell lymphoma, suggesting tissue-specific function of NuRD (Loughran, S.J., et al., J Exp Med, 2017). Interestingly, in our present study we have identified for the first time that endogenous UTX, but not JMJD3, reversibly co-immunoprecipitates with NuRD in AML cells. These findings led us to test the hypothesis whether UTX would participate with NuRD in AML. ChIP-seq analysis in AML blasts using antibodies against UTX and CHD4 (intact ATPase component of leukemic NuRD) along with H3K27ac identified the co-localized genes. ChIP-qPCR, transcriptome, pathway analysis (P<0.001) performed in paired AML, and MBD3loss of function experiments suggested an enrichment of Dedicator of Cytokinesis (DOCK) transcripts as bona fide effectors of UTX and NuRD in AML.
DOCK proteins are conserved atypical guanine nucleotide exchange factors (GEFs) for Rho GTPase activation, regulating cell motility and invasion. Earlier we had shown that small GTPases regulate myeloid leukemia cell engraftment, survival in vivo (Sengupta A et al., Blood, 2010). DOCK1 upregulation is associated with a poor prognosis in AML (Hwei, L.S., Blood, 2016). TCGA cross-cancer analysis showed that UTX is maximally expressed, whereas MBD3 is downregulated in AML among all cancer types. Consistent with this observation, DOCK expression was significantly (P<0.001) increased in MBD3loUTXhi AML cohort compared to MBD3hiUTXlo AML. Importantly, MBD3loUTXhi patients have relatively poor survival compared to MBD3hiUTXhi individuals, indicating that a combination of high UTX and low MBD3 expression could be a marker of poor prognosis in AML.
Mechanistically, MBD3 deficiency caused loss of HDAC1 occupancy with a corresponding increase in UTX, CBP and H3K27ac on target DOCK loci leading to de-repression of gene expression. In agreement with this finding, loss of MBD3 resulted in ~ 2-fold increase in active Rac GTP and promoted AML cell migration to CXCL12. Interestingly, UTX silencing opposed DOCK expression, Rac activation and reversed hyper-migratory phenotype of MBD3-deficient AML cells. Together, these data account for UTX and MBD3 epistasis in regulating DOCK-Rac signalling in AML. Finally, treatment with DOCK inhibitor CPYPP dramatically inhibited survival of AML cells while having minimal effect on the survival of normal CD34+ cells. In unison, our findings highlight UTX as a putative oncogene in conjunction with leukemic NuRDand posit DOCK proteins as an important target of UTX-NuRD axis in human AML cells.
To conclude, we provide evidence for MBD3-deficient NuRD in leukemia pathobiology, and inform a novel epistasis between UTX and NuRD towards maintenance of oncogenic gene expression in AML, and rationalize DOCK inhibition as a novel therapeutic modality for precision medicine in AML.
No relevant conflicts of interest to declare.
Asterisk with author names denotes non-ASH members.