The Histone Methyltransferase DOT1L Cooperates with LSD1 to Control Cell Division in Blast-Phase MPN

The establishment of strategies to reduce the fitness of pathogenic clones and diminish disease burden remains a critical challenge in the development of novel therapies for Myeloproliferative Neoplasms (MPN). While JAK inhibitors control inflammation and hyperproliferation, JAK2-mutated clones persist and may undergo clonal evolution and eventually malignant transformation. Aside from allogenic stem cell transplantation, none of the established treatment regimens for MPN consistently allows control of the malignant clone, particularly in high-risk patients experiencing progression to blast-phase MPN. Given the advanced age of many of those patients and the unfavorable outcomes of blast-phase MPN novel therapeutic approaches are urgently needed.

Bomedemstat, a small molecule inhibitor of the histone demethylase LSD1 (KDM1A), has recently evolved as a promising new therapeutic tool in MPN. In clinical trials Bomedemstat showed disease modifying activity reducing both MPN symptoms and clonal burden in a significant number of patients. Whether LSD1-inhibitors retain their ability to modify clonal fitness during progression to blast-phase remains unknown.

We aimed to identify cooperative epigenetic modifiers using a chromatin-focused CRISPR-Cas9 screen. Therefore, we transduced Human Erythroleukemia (HEL) cells, a model for JAK2-mutated blast-phase MPN, with Cas9 and a library of 7856 sgRNAs targeting 719 chromatin-related genes. The cells were then treated with Bomedemstat (100nM), the LSD1-inhibitor GSK2879552 (100nM) or DMSO as diluent control. Of note, the histone methyltransferase DOT1L showed strong synthetic lethal activity with both LSD1-inhibitors. DOT1L catalyzes histone 3 lysine 79 methylation (H3K79me) at transcriptionally active genes and has been shown to be an essential component of an oncogenic chromatin complex in MLL (KMT2A)-rearranged acute leukemia. Hence, we aimed to define the role of DOT1L in blast-phase MPN. CRISPR-Cas9 mediated knockout of DOT1L led to a significant reduction in cellular fitness of HEL cells reflected by a reduced proliferation rate (doubling times: 30h in WT vs. 49h in DOT1L-ko). Furthermore, NXG mice transplanted with DOT1L-ko cells showed a prolonged survival (median survival: 107d for DOT1L-ko (n=7) vs. NT-control (n=4), p<0.001) and a reduction of disease penetrance with 40% of animals not developing blast-phase MPN. Interestingly, the genes regulated by DOT1L in HEL cells differ from those in MLL-rearranged AML, including several JAK-signaling targets (e.g. MPL, JAK1, DUSP6, YBX1, CDK4, CDK6, MYC).

To validate the synthetic lethal interaction between DOT1L and LSD1 we treated DOT1L-ko cells with Bomedemstat or GSK2879552 and discovered a 100-fold increase in drug sensitivity compared to WT cells (MTS-assay). This cooperative effect was caused by an increase in apoptosis (Annexin V, DOT1L-ko+DMSO: 23% vs. DOT1L-ko+Bomedemstat: 39%, p=0.003) and a blockade of S-phase entry (EDU-assay, DOT1L-ko+DMSO: 37.6% vs. DOT1L-ko+Bomedemstat: 8.5%, p<0.001). In RNAseq, LSD1-inhibitor treatment led to a strong repression of cell cycle mediators and an induction of apoptosis-related genes only in DOT1L-ko cells (DESeq2, padj<0.05, FC>2). ChiPseq revealed that most regions at which LSD1 was bound, were co-occupied by DOT1L. Those regions were marked with H3K4me1 and had accessible chromatin indicating LSD1-binding to enhancers. After knockout of DOT1L, chromatin accessibility at those enhancers decreased significantly demonstrating a so far unrecognized role of DOT1L in controlling enhancer activation. Strikingly, most of the LSD1-DOT1L co-occupied enhancers lacked H3K79me2 indicating that DOT1L’s function at these sites is independent of its enzymatic activity. Consistent with this observation, treatment of the MPN blast-phase cell lines HEL or SET2 with LSD1-inhibitors in combination with the DOT1L-inhibitor EPZ5676 only showed a modest cooperative effect.

In summary, our work demonstrates functional cooperativity between DOT1L and LSD1 in blast-phase MPN. This cooperation is caused by orchestrated binding of DOT1L and LSD1 at selected enhancer regions and is independent of DOT1L’s enzymatic activity. This non-canonical function of DOT1L in blast-phase MPN provides a strong rationale for the development of targeted protein degraders (PROTACs) of DOT1L to exploit these findings therapeutically.