2017 – ARID1B-DEPENDENT ENHANCER REWIRING SUSTAINS KMT2A-MLLT3 LEUKEMIA AND EXPOSES A THERAPEUTIC VULNERABILITY

Abstract

KMT2A-rearranged acute myeloid leukemia (AML), one of the most treatment-refractory leukemic subtypes, is driven by fusion protein-mediated promoter hijacking of self-renewal genes such as HOXA9 and MEIS1. However, how these direct targets promote genome-wide transcriptional reprogramming remains incompletely resolved. Using a doxycycline-inducible KMT2A-MLLT3 mouse AML model combined with epigenomic profiling (ATAC-seq, ChIP-seq, promoter capture Hi-C), we demonstrate that KMT2A-MLLT3 supports the establishment of a two-phase oncogenic circuit by directly driving the expression of a set of core target genes, including HOXA9, MEIS1, and ARID1B, which subsequently unleashes a major remodeling of the enhancer landscape to sustain leukemogenesis. Pooled CRISPR knockout screens confirmed the functional requirement of these leukemia-activated enhancers for leukemic growth. We identify ARID1B, a subunit of the cBAF chromatin remodeling complex, as a key mediator of enhancer remodeling. Specifically, ARID1B supports accessibility at leukemia-activated enhancers, and its loss selectively collapses the enhancer program and impairs leukemic, but not normal hematopoietic, growth, across species. Clinically, high ARID1B expression correlates with earlier disease onset and leukemic stem cell transcriptional signatures. We further demonstrate that strengthening of the ARID1B-driven enhancer network is associated with BET inhibitor resistance, which can be overcome by combined cBAF and BET inhibition. Thus, our work identifies ARID1B-mediated enhancer remodeling as a key oncogenic driver in KMT2A-rearranged AML with potential clinical implications.

KMT2A-rearranged acute myeloid leukemia (AML), one of the most treatment-refractory leukemic subtypes, is driven by fusion protein-mediated promoter hijacking of self-renewal genes such as HOXA9 and MEIS1. However, how these direct targets promote genome-wide transcriptional reprogramming remains incompletely resolved. Using a doxycycline-inducible KMT2A-MLLT3 mouse AML model combined with epigenomic profiling (ATAC-seq, ChIP-seq, promoter capture Hi-C), we demonstrate that KMT2A-MLLT3 supports the establishment of a two-phase oncogenic circuit by directly driving the expression of a set of core target genes, including HOXA9, MEIS1, and ARID1B, which subsequently unleashes a major remodeling of the enhancer landscape to sustain leukemogenesis. Pooled CRISPR knockout screens confirmed the functional requirement of these leukemia-activated enhancers for leukemic growth. We identify ARID1B, a subunit of the cBAF chromatin remodeling complex, as a key mediator of enhancer remodeling. Specifically, ARID1B supports accessibility at leukemia-activated enhancers, and its loss selectively collapses the enhancer program and impairs leukemic, but not normal hematopoietic, growth, across species. Clinically, high ARID1B expression correlates with earlier disease onset and leukemic stem cell transcriptional signatures. We further demonstrate that strengthening of the ARID1B-driven enhancer network is associated with BET inhibitor resistance, which can be overcome by combined cBAF and BET inhibition. Thus, our work identifies ARID1B-mediated enhancer remodeling as a key oncogenic driver in KMT2A-rearranged AML with potential clinical implications.