Research Article: HoxA9 binds and represses the Cebpa +8 kb enhancer

Date Published: May 23, 2019

Publisher: Public Library of Science

Author(s): Lei Peng, Hong Guo, Peilin Ma, Yuqing Sun, Lauren Dennison, Peter D. Aplan, Jay L. Hess, Alan D. Friedman, Roberto Mantovani.


C/EBPα plays a key role in specifying myeloid lineage development. HoxA9 is expressed in myeloid progenitors, with its level diminishing during myeloid maturation, and HOXA9 is over-expressed in a majority of acute myeloid leukemia cases, including those expressing NUP98-HOXD13. The objective of this study was to determine whether HoxA9 directly represses Cebpa gene expression. We find 4-fold increased HoxA9 and 5-fold reduced Cebpa in marrow common myeloid and LSK progenitors from Vav-NUP98-HOXD13 transgenic mice. Conversely, HoxA9 decreases 5-fold while Cebpa increases during granulocytic differentiation of 32Dcl3 myeloid cells. Activation of exogenous HoxA9-ER in 32Dcl3 cells reduces Cebpa mRNA even in the presence of cycloheximide, suggesting direct repression. Cebpa transcription in murine myeloid cells is regulated by a hematopoietic-specific +37 kb enhancer and by a more widely active +8 kb enhancer. ChIP-Seq analysis of primary myeloid progenitor cells expressing exogenous HoxA9 or HoxA9-ER demonstrates that HoxA9 localizes to both the +8 kb and +37 kb Cebpa enhancers. Gel shift analysis demonstrates HoxA9 binding to three consensus sites in the +8 kb enhancer, but no affinity for the single near-consensus site present in the +37 kb enhancer. Activity of a Cebpa +8 kb enhancer/promoter-luciferase reporter in 32Dcl3 or MOLM14 myeloid cells is increased ~2-fold by mutation of its three HOXA9-binding sites, suggesting that endogenous HoxA9 represses +8 kb Cebpa enhancer activity. In contrast, mutation of five C/EBPα-binding sites in the +8 kb enhancer reduces activity 3-fold. Finally, expression of a +37 kb enhancer/promoter-hCD4 transgene reporter is reduced ~2-fold in marrow common myeloid progenitors when the Vav-NUP98-HOXD13 transgene is introduced. Overall, these data support the conclusion that HoxA9 represses Cebpa expression, at least in part via inhibition of its +8 kb enhancer, potentially allowing normal myeloid progenitors to maintain immaturity and contributing to the pathogenesis of acute myeloid leukemia associated with increased HOXA9.

Partial Text

Hox proteins are best known to mediate pattern formation during early development, but a subset serve additional functions in adult tissues. HoxA9 is preferentially expressed in myeloid progenitors during hematopoiesis, and its level diminishes during normal myeloid maturation [1–3]. Notably, HOXA9 is over-expressed up to 13-fold in >50% of acute myeloid leukemia (AML) cases, and its increased expression is associated with poor prognosis [4, 5]. Golub et al 1999 found that of 6,187 genes evaluated, HOXA9 over-expression was most highly correlated with treatment failure. Andreef et al 2009 evaluated 119 adult AML cases and that found 20% and 10% long-term survival amongst patients with intermediate or high levels of HOXA9, respectively, compared with 40% survival amongst patients with low-level HOXA9; in addition, they noted that patients with low-level HOXA9 mainly had favorable cytogenetics, i.e. t(15;17), t(8;21), or inv(16). HOXA9 gene expression has been found to be up-regulated in AML cases as a consequence of gene activation by MLL fusion proteins, NUP98 fusion proteins, CALM-AF10, NPM1c mutation, or decreased EZH2 or ASXL1, each often associated with intermediate- or high-risk cases [6].

C/EBPα is required for normal myeloid development, and reduced C/EBPα expression or activity is evident in the majority of AML cases [11]. The murine Cebpa gene contains a 455 bp enhancer located at +37 kb that is highly conserved in the human CEBPA locus and functions specifically in hematopoietic stem and progenitor cells [12–15]. The +37 kb enhancer is bound and activated by RUNX1, C/EBPα, GATA-2, SCL, NR4A1/3, PU.1, additional ETS family transcription factors, and potentially c-Myb in normal myeloid progenitors and is bound and potentially repressed by RUNX1-ETO in AML cases [13, 27, 28]. Analysis of H3K4me1 and H3K27Ac histone marks in murine GMP, CMP, and additional marrow hematopoietic progenitors identified a second potential enhancer located at +8 kb in the Cebpa locus [13]. Similar analysis of H3K27Ac marks in the human CEBPA locus indicates that the corresponding +9 kb enhancer is active in myeloid cells and also in all non-myeloid, C/EBPα-expressing tissues examined, i.e. liver, lung, adipose, large and small intestines, skin epithelium, and placenta; moreover, 4C-Seq analysis demonstrates interaction of the +9 kb enhancer with the CEBPA promoter in both myeloid and non-myeloid cell lines [15]. Our prior analysis of ChIP-Seq data indicates that C/EBPα, but not RUNX1, PU.1, GATA-2, or SCL, binds the +8 kb enhancer [13]. We now further demonstrate that HoxA9 directly binds the +8 kb Cebpa enhancer via three binding sites and that HoxA9 also localizes to the +37 kb enhancer, likely via interaction with other bound transcription factors. We show that increased HoxA9 resulting from NHD13 myeloid oncoprotein expression is associated with reduced Cebpa in CMP and LSK marrow cells, decreased HoxA9 as 32Dcl3 cells differentiate is associated with increased Cebpa, activation of exogenous HoxA9-ER reduces Cebpa even in the presence of a ribosomal translation inhibitor, and mutation of the three HoxA9-binding sites in the +8 kb enhancer increases reporter activity in two myeloid cell lines. Moreover, inactivation of HoxA9-ER in myeloid progenitors leads to increased Cebpa mRNA [8]. Together, these data indicate that HoxA9 binds and represses Cebpa +8 kb enhancer activity.