Loss-of-function mutations within KMT2D are a striking feature of the germinal centre lymphomas, with lesions present in 80% of Follicular Lymphoma (FL) and 30% of Diffuse Large B-cell Lymphoma (DLBCL) cases, resulting in decreased H3K4 methylation and altered gene expression. The majority of KMT2D mutations in FL and DLBCL are truncating and clonal, arise early during tumour development, and are often bi-allelic, yet despite their frequency, no attempts have been made to therapeutically target these mutations. We hypothesised that inhibition of the KDM5 family, which normally opposes KMT2D through demethylating H3K4me3/2, would re-establish H3K4 methylation and restore the expression of genes repressed upon loss of KMT2D.
KDM5-inh1 is a selective inhibitor of the KDM5 family developed by EpiTherapeutics and Gilead. The aim of this study was to determine whether inhibition of KDM5 by KDM5-inh1 could counteract the effects of KMT2D mutations and provide a potential targeted therapy.
The in vitro efficacy of KDM5 inhibition (KDM5i) was assessed over a panel of genetically characterised lymphoma cell lines (eight KMT2D mutant, six KMT2D WT), whilst in vivo efficacy was determined in a SU-DHL-6 xenograft model. CRISPR was used to generate isogenic cell lines where KMT2D was deleted in WT WSU-DLCL2 cells and corrected in mutant SU-DHL-8 cells. KDM5 target genes were identified through ChIP- and RNA-seq analysis.
Exposure of lymphoma cells to KDM5-inh1 increased global levels of the promoter associated H3K4me3 histone mark and subtly decreased levels of enhancer associated H3K4me1. After five days, we observed an in vitro anti-proliferative and cytostatic response that was markedly greater in both endogenous and CRISPR-edited KMT2D mutant DLBCL cell lines, whilst KDM5-inh1 also inhibited tumour growth in KMT2D mutant xenografts in vivo. To identify key downstream mediators of KDM5i, we performed ChIP- and RNA-seq analysis in KDM5i sensitive and insensitive cell lines, and identified 2408 regions with altered H3K4me3 levels, with 98% displaying increased levels of H3K4me3 and 62% having previously been identified to be bound by KMT2D. KDM5i target regions overlapped with both enhancers and promoters however the increases in promoter H3K4me3 appeared to drive gene expression changes to a greater degree than those at enhancers. Mechanistically, we have been able to identify two pathways through which KDM5i function. Firstly, KMD5i increased the expression of a number of KMT2D-regulated B cell signalling repressors (e.g. PTPN6, FCGR2B, FCRL5), resulting in diminished B-cell receptor signalling following anti-IgM stimulation in KDM5i sensitive cells. Secondly, KDM5i results in a significant alteration in the expression of BCL2 family members (BCL2, BCL-XL, BIM, NOXA), with BCL2 downregulated in every t(14;18) positive cell line and primary FL cell suspension examined.
We detail a novel way of reverting the effects of loss-of-function mutations in the histone methyltransferase KMT2D by inhibiting the corresponding KDM5 demethylase family, stabilising levels of H3K4 methylation and restoring expression of KMT2D regulated genes. KDM5i may therefore offer an effective therapeutic strategy for targeting KMT2D mutations in malignancies such as germinal centre lymphomas.