Histone Profiling of Normal B-Precursors and Primary Pre-B Acute Lymphoblastic Leukemia Reveals Distinct Aberrant Histone Codes In MLL-Rearranged Vs. Wild Type MLL Leukemias That Correlate with Differential Expression of Key MLL Target Genes

Epigenetic regulation of gene transcription is mediated both by methylation of DNA CpG islands and the local configuration of chromatin, which is dynamically regulated by post-translational modifications, or “marks”, of key lysines (K) of histones (especially H3). Some marks are associated with transcriptional repression [trimethylation (me3) of K9 and K27], and some with activation [me3 of K4, dimethylation (me2) of K79 and acetylation (Ac) of K9 and K14]. The MLL gene encodes a protein that functions as a master regulator of target gene expression by methylating H3K4 via its SET domain, and by interacting with other proteins with histone modifying properties. MLL is frequently rearranged (MLL-r) by translocations in acute leukemias, which exhibit a distinct global gene expression pattern. Many MLL-r partner genes form complexes that can methylate H3K79. Thus, histone modifications may be central to the function of both wild type (MLL-wt) and MLL-r. We hypothesized that aberrant histone coding of target genes contributes to MLL-r leukemogenesis. We characterized the histone code associated with the promoters of selected genes in n=5 MLL-r pre-B ALL samples (MLL-AF4 or MLL-ENL), n=4 MLL-wt pre-B ALL samples (TEL-AML1 or hyperdiploid) and normal control B-precursors (CD19+ cord blood cells). We selected 9 genes differentially overexpressed in MLL-r leukemia (HOXA7, HOXA9, MEIS1, FLT3, CCNA1, ZC3H12C, ATP8B4, C20orf103, and PROM1), and 3 control genes that are not MLL targets (HOXA1, HOXC8, LTF). We performed ChIP with antibodies specific for key H3 modifications (K4me3, K9me3, K9/14Ac, K27me3 and K79me2), followed by qPCR for the selected genes. Expression was measured by RT/qPCR. All 9 MLL target genes were significantly overexpressed in the MLL-r cohort, and this was associated with a specific “activating” histone code at the genes' promoters (fig 1 – MEIS1, e.g.). The opposite “repressive” code was found in the MLL-wt cohort, and in the MLL-r cohort at the promoters of the control genes. Compared to both sets of leukemias, normal B-precursors exhibited a paucity of histone modifications for all genes. For most genes, a specific developmental pattern of alterations in the histone code and corresponding relative change in expression could be traced from the normal B-precursors to the leukemia cells. This pattern was strikingly different in MLL-r leukemias than in MLL-wt leukemias, suggesting that the acquisition of MLL-r by normal B-precursors causes altered gene expression patterns via changes in the histone code. For most genes, normal B-precursors exhibit both the activating K4me3 mark and the repressive K27me3 mark, and express low but detectable levels of RNA. In MLL-r leukemias, upregulation of genes is associated with an increase in K4me3, loss of K27me3, and gain of K9/14Ac and/or K79me2. In MLL-wt leukemias, silencing of genes is associated with loss of K4me3 and gain of K9me3. To study the direct effects of MLL-wt and MLL-r on the histone code, we used 2 rounds of siRNA over 48 hours to knock down MLL-AF4 only, MLL-wt only or both in the RS4;11 cell line (MLL-AF4+ B-precursor ALL), then performed RT/qPCR and ChIP/qPCR. We achieved at least 60% knock down of MLL-AF4 and/or MLL-wt. Knock down of MLL-wt, with or without concomitant knock down of MLL-AF4, did not diminish the K4me3 mark for any genes, suggesting that MLL's SET domain is not required to maintain K4 methylation. While knock down of MLL-AF4 or MLL-wt alone did not diminish K79 methylation, knock down of both completely removed the K79me2 mark from all genes, suggesting that expression of either MLL-wt or MLL-AF4 is absolutely required for H3K79 methyltransferase activity. Two genes (HOXA7 and PROM1) demonstrated evidence of direct transcriptional regulation by MLL-AF4, since their expression decreased markedly after knock down of MLL-AF4 alone or with MLL-wt, but not with MLL-wt alone. In summary, primary MLL-r pre-B ALLs exhibit a distinct activating histone code at key overexpressed target genes when compared to MLL-wt pre-B ALLs and normal B-precursors. A causative role for MLL fusion proteins is suggested by the distinct pattern of histone code progression from normal B-precursors to MLL-r leukemias. Furthermore, knock down experiments provide direct evidence that some of the observed histone modifications in MLL-r leukemia, particularly H3K79 methylation, are directly downstream of wild type and mutant MLL.