CEBPα Is a Transcriptional Repressor of T-Cell Related Genes Explaining the Myeloid/T-Lymphoid Features of CEBPα-Silenced AML

Abstract 554

Acute myeloid leukemia (AML) is a heterogeneous disease of different subtypes characterized by distinct cytogenetic or molecular abnormalities. We recently identified a previously unrecognized subtype with a unique epigenetic feature, i.e. silencing of the gene that encodes CCAAT-enhancer binding protein alpha (CEBPα) by DNA hypermethylation (denoted as the CEBPα^silenced group). The leukemic blast cells of these patients express myeloid as well as T-lymphoid markers. Moreover, gene expression and DNA methylation profiling put these leukemias in between AML and T-lymphoid leukemia (T-ALL). CEBPα is a transcription regulator that is essential for normal neutrophil development. We hypothesize that methylation and consequently silencing of the gene encoding CEBPα is abnormal and an important hit in the transformation of this unique leukemia subtype. The mechanism by which CEBPα silencing plays a role in transformation of cells with myeloid/T-lymphoid features is the aim of our study. We carried out gene expression profiling of the CEBPα^silenced group (n=10) and compared gene expression levels to normal CD34⁺ bone marrow cells (n=11) and the remaining AML group (n=506) using a three-way ANOVA and a post-hoc test (tukey-kramer method). We detected 686 differentially expressed genes with P < 0.05 after multiple testing. Of these 686 genes, 288 were up- and 401 down-regulated in the CEBPα^silenced group. We next asked the question which of those genes might be bona fide CEBPα targets and whether expression has been altered as the result of CEBPα silencing. We transduced estrogen-inducible C/EBPα in 32D cells and carried out ChIP-on-chip analysis using ER specific antibodies. The analysis yielded a collection of 529 CEBPα target genes that are significant enriched for C/EBPα binding (P < 0.05) using Hypergeometric Analysis of Tiling-arrays (HAT). We hypothesized that the direct target genes of CEBPα, derived from the 32D model system, were also present among deregulated genes in CEBPα^silenced human AMLs. We therefore overlaid the detected direct binding targets of CEBPα in 32D cells, with the differentially expressed genes in the CEBPα^silenced group and identified 49 overlapping genes (P=1×10⁻⁷) as putative direct targets. Among these 49 genes, 25 were down-regulated and 24 were upregulated in the primary CEBPα^silenced AML group. Both groups of genes were highly enriched for the CEBPα consensus binding sequence, i.e. 16/25 and 20/24 promoter regions respectively. The 25 downregulated genes, among which ACSL1, MYCT1 or Slc7a11, represent targets that are most likely normally activated by CEBPα, but are not transcribed in CEBPα^silenced human AMLs due to the absence of CEBPα. Among the genes that were upregulated in CEBPα^silenced leukemias are BCL2, CCR9, CEBPG or CD47. These putative target genes are repressed in CEBPα expressing cells and activated when CEBPα is silenced. This observation suggests that the transcription factor CEBPα may also acts as a repressor of gene transcription. We therefore studied the expression of two of those genes, i.e. CEBPG and CCR9 in Lin-Sca+Kit+ (LSK) bone marrow hematopoietic stem cells from wild type versus conditional Mx-Cre/CEBPα knock-out mice. Similar to what we observed in human CEBPα^silenced leukemias, we found that these two genes were switched off in LSK cells from CEBPα knock-out animals. Ingenuity pathway analysis of the 24 upregulated genes detected high enrichment (P < 0.001) of pathways involved with T-cell development. Our data clearly suggest that CEBPα may act as repressor of T-cell related genes through direct promoter interaction. We therefore propose that silencing of CEBPα by promoter hypermethylation is one of the transforming events that driving towards mixed myeloid/T-lymphoid leukemia.