The genomic alterations that drive acute myeloid leukemia (AML) are well established, however, the molecular pathways that are deregulated as a consequence of these mutations and their impact on AML development and progression remain poorly understood. We and others have previously shown that the transcription factor (TF) JUN, though not mutated, is highly expressed in multiple AML sub-types (e.g. normal karyotype (NK-AML), complex karyotype and 11q23 rearrangements) relative to healthy hematopoietic stem and progenitor cells (HSPCs). We also recently reported that JUN supports cell survival in a panel of genetically diverse AML cell lines and patient-derived AML samples ex vivo. Moreover, we showed that JUN inhibition significantly delays disease onset and improves survival in a genetically engineered mouse model (GEMM) of AML driven by MLL-AF9. Of critical importance, we have now observed that deletion of JUN does not negatively impact steady-state hematopoiesis or the hematopoietic reconstituting potential of HSPCs. While these results position JUN as an attractive therapeutic target in AML, previous efforts to develop JUN inhibitors have been unsuccessful. Therefore, our objective is to elucidate the mechanisms by which JUN supports AML in order to identify putative therapeutic targets in this pathway.

JUN regulates many transcriptional programs by forming hetero-dimers with one of many other TFs and our goal was to identify which dimeric partner cooperates with JUN to regulate transcriptional programs that support AML. We previously reported that JUN supports AML, in part, by regulating the transcriptional output of the unfolded protein response (UPR), which is a signal transduction network that helps cells negotiate endoplasmic reticulum (ER) stress. To identify dimerization partners that cooperate with JUN to drive transcriptional output of the UPR, we utilized the UCSC genome browser, which contains the chromatin immunoprecipitation-sequencing (ChIP-seq) data for JUN as well as 15 known dimerization partners of JUN. Using this database, we found that JUN localizes to the promoters of 35 UPR genes. Notably, we found that the stress-response factor ATF3 co-localizes with JUN at 31 of these 35 promoters (88.6%), exceeding the co-localization of the other 14 factors. Consistent with these observations, we found that shRNA-mediated inhibition of either JUN or ATF3 leads to decreased mRNA expression of the UPR targets genes (XBP1s, ATF4, DDIT3, HSPA5, GRP94, ATF6, EIF2AK3and ERN1) in human AML cell lines (e.g. THP-1, OCI-AML3, U937, MOLM14) as well as leukemia cells derived from the MLL-AF9-driven GEMM of AML. We also observed that inhibition of either JUN or ATF3 blocked the induction of UPR signaling and rendered human and mouse AML cells more susceptible to ER stress-induced death further suggesting that JUN and ATF3 regulate the transcriptional output of the UPR.

To determine if ATF3 expression, like JUN, is deregulated in human AML, we performed a retrospective analysis of AML patient gene expression profiles and found that elevated expression of ATF3 is significantly associated with worse outcomes in two distinct datasets from NK-AML patients (p=0.0248, p=0.0049). Furthermore, we found that ATF3 expression is higher in AMLs with FLT3-ITD or 11q23 rearrangements, compared to healthy HSPCs. Based on these data, we investigated the functional role of ATF3 in AML. First, we found that shRNA-mediated inhibition of ATF3 significantly reduced the colony forming capacity of mouse AML cells expressing MLL-AF9 but, importantly, not healthy HSPCs. Furthermore, we found that ATF3 inhibition impeded the growth and survival of AML cells from a second GEMM of AML driven by the deletion of Dnmt3a and Tet2 in combination with expression of FLT3-ITD. We also observed that, similar to JUN inhibition, shRNA-mediated inhibition of ATF3 significantly reduced disease burden in MLL-AF9-driven AML in vivo (p=0.0028). At the cellular level, we found that AML cells depleted of ATF3 displayed increased CD11b expression and annexin V staining suggesting that ATF3 may support the differentiation blockade in AML.

Collectively, our data suggest that JUN and ATF3 cooperate to support a variety of AML subtypes by regulating transcriptional programs, such as the UPR and our current efforts are focused on identifying additional targetable JUN/ATF3-regulated transcriptional programs that support AML.


No relevant conflicts of interest to declare.

Author notes


Asterisk with author names denotes non-ASH members.