Page 239 - 2021_09-Haematologica-web

Letters to the Editor
Targeting BRD4 in acute myeloid leukemia with partial tandem duplication of the MLL gene
The lysine methyltransferase 2a (KMT2A) (which is also known and hereafter referred to as mixed-lineage leukemia [MLL], trithorax [Drosophila] homolog gene) plays a pivotal role in embryogenesis and hematopoiesis. Recurrent, balanced translocations involving the MLL gene [t(v;11q23)] are heterogeneous, and more than 75 different fusion partners have been described as impor- tant drivers in acute myeloid leukemia (AML) leukemo- genesis.1,2 Beside chromosomal aberrations, a unique gene rearrangement in MLL known as partial tandem duplication (PTD) can be found in approximately 5-11% of cytogenetically normal AML (CN-AML) patients. This mutation is associated with poor prognosis.3-8 Although both t(v;11q23) and the MLL-PTD result in increased HOMEOBOX (HOX) gene expression in leukemic blasts, t(v;11q23) have been shown to be genetically and func- tionally distinct from the MLL-PTD.9 While almost all t(v;11q23) lose their C-terminal transactivation and methyltransferase domains, these C-terminal domains are retained in the MLL-PTD.9
The transcription factor BRD4 is a member of the bro- modomain and extra terminal (BET) family of proteins. Aberrant BRD4 binding and gene activation has been shown to be important for t(v;11q23)-mediated leukemo- genesis.10 JQ1 is one of the best-characterized, small mol- ecule bromodomain inhibitors.11 However, the potential use of JQ1 in MLL-PTD AML has not been extensively studied yet. Therefore, we examined whether MLL-PTD AML blasts are sensitive to JQ1 treatment and if BRD4 inhibition results in an altered binding of the transcrip- tion factor to DNA.12
First, we tested whether JQ1 treatment has an impact on cell proliferation and survival in a MLL-PTD+ AML cell line (i.e., EOL-1) and in a MLL wild-type cell line (i.e., K562).13 Both cell lines were treated with JQ1 or dimethyl sulfoxide (DMSO) vehicle control for 24 hours (h) at dif- ferent concentrations and cell growth was assessed by WST-1 assay. We found a significant decrease in cell pro- liferation in EOL1 cells (IC50 = 321 nM) but not in K562 cells (Figure 1A). Concomitantly, we found a significant and dose dependent increase in the number of apoptotic EOL-1 cells (Figure 1B) but not in the K562 cells (Online Supplementary Figure S1A). We also analyzed the effect of JQ1 treatment on primary blast cell growth from three AML patients that harbor a MLL-PTD compared to nor- mal hematopoietic stem and progenitor cells (HSPC; CD34+ cord blood) controls. We found JQ1 treatment sig- nificantly reduced blast cell growth assessed by a decrease in the number of colony-forming cells (CFC) in JQ1-treated MLL-PTD AML samples, with no significant decreases of CFC in normal HSPC (Figure 1C) or MLL wild-type (wt) primary patient samples (Online Supplementary Figure S1B).
Next, we tested the effect of JQ1 in a murine AML mouse model. For these experiments we used our well established MllPTD/WT Flt3ITD/WT double knockin AML mouse model14,15 that develops lethal CN-AML with ~100% penetrance. In secondary bone marrow trans- plantation, it leads to death within 6 to 12 weeks.14,15 Of note, MLL-PTD is predominantly found in CN-AML in humans. First, we wanted to determine whether JQ1 also induced apoptosis in the MllPTD/WT Flt3ITD/WT mouse AML blasts, similar to what we observed in human AML blasts. We found that JQ1 induced a significant increase in apoptosis assessed by Annexin V staining in the
MllPTD/WT Flt3ITD/WT blasts (Figure 1D) with essentially no toxicity to normal murine bone marrow cells (Online Supplementary Figure S1C). Based on these promising in vitro results, we then wanted to test, whether targeting BRD4 in vivo would result in prolonged survival of mice with Mll-PTD+ leukemia. In order to test the anti- leukemic activity of JQ1 in a murine AML model, we used our previously established MllPTD/WT Flt3ITD/WT mouse model.14-16 We observed a significant increase in survival of JQ1 treated MllPTD/WT Flt3ITD/WT mice compared to mice treated with vehicle control (Figure 1E). Moreover, the mice that eventually succumbed to disease and were treated with JQ1 had significant lower spleen weight, indicating a lower leukemic burden (Figure 1F). Interestingly, we also found that the leukemic bone mar- row cells from JQ1-treated mice had a significant lower engraftment potential after re-transplantation than cells from mice treated with vehicle control (Online Supplementary Figure S1D). These data suggested that JQ1 might also have an impact on leukemic cell self-renewal and consequently leukemia stem cells (LSC) in Mll-PTD AML, however further experiments are needed to fully address effects on LSC by JQ1 in Mll-PTD leukemia.
After identifying the ability of JQ1 to decrease MLL-PTD+ AML blast growth in vitro and in vivo, we wanted to determine whether alterations in BRD4 bind- ing accounts for MLL-PTD blast sensitivity to BRD4-inhi- bition. BRD4 has been shown to be a positive regulator of gene transcription and aberrant BRD4-binding in can- cer induces alterations in gene expression. Thus, we hypothesize that MLL-PTD leukemogenesis is driven by dysregulation of gene expression patterns resulting from aberrant BRD4 binding. Furthermore, we wanted to determine whether normal BRD4 binding could be restored by treatment with JQ1. In order to address this question, we performed total RNA sequencing (RNA-seq) on primary MLL-PTD AML blasts and normal HSPC treated with JQ1 or vehicle control (n=3 for each group, pooled) before and after JQ1 treatment and analyzed as previously described.18 In addition, we performed chro- matin immunoprecipitation sequencing (ChIP-seq) using a BRD4 antibody. Cells from three healthy cord blood donors (CB) and primary leukemic cells from three patients with MLL-PTD were treated with either JQ1 (10 nM) or vehicle (DMSO) for 24 hours (h) and ChIP was performed as previously described.18 The 75-basepair sequence reads were generated using an llumina sequencing platform (NextSeq 500) and then mapped to the human reference genome (GRCh37/hg19) using the BWA algorithm with default settings.19 Aligned reads were normalized and genomic regions with local enrich- ments against corresponding input sample, peaks, were defined using MACS algorithm with a cutoff P-value of 1e-7.20 Consensus peaks were defined by merging over- lapping peak coordinates and peak scores were calculat- ed. The resulting matrix was annotated with gene infor- mation by calculating distances from RefSeq gene starts and ends to the center of the consensus peak regions and applying an annotation cutoff of 5 kb. Similar to what has previously been described for t(v;11q23)-AML that BRD4 has an aberrant binding profile, we found three times more and stronger genomic interactions of BRD4 in the MLL-PTD patient sample compared to normal HSPC, and only 22% of the BRD4 peaks in MLL-PTD over- lapped with normal HSPC in ChIP-seq (Figure 2A). When we treated the MLL-PTD sample with JQ1, both the number of peaks and their intensity decreased, whereas JQ1 treatment did not affect BRD4 binding in HSPC (Figure 2B). Next, by integrating the RNA-seq and ChIP-
haematologica | 2021; 106(9)
2527