Background: Decitabine (DAC) is one of the first generation of epigenetic drugs used clinically to treat myelodysplastic syndromes (MDS) and acute myeloid leukemia (AML). Historically, DAC was considered as a cytotoxic agent when it was first developed in the 1960s. Subsequent studies in 1980s showed that DAC can induce DNMT1 degradation and DNA hypomethylation at non-cytotoxic concentration. Since then, DNA demethylation has been considered as the main mechanism of action of DAC. The DAC-induced DNA hypomethylation was shown to induce upregulation of tumor suppressor genes or endogenous retroviruses, which could contribute to the antitumor effect of DAC. Recently, DAC was also shown to be effective even in MDS and AML with TP53 mutations. However, it remains unclear why myeloid tumors, especially those with TP53 mutations, are particularly susceptible to DAC treatment. Methods: Cancer Dependency Map (DepMap) database was used to evaluate in vitro sensitivity of human cancer cell lines to DAC. To identify essential genes that regulate sensitivity of myeloid tumors to DAC, we performed a genome-wide CRISPR-dCas9 activation screen using murine MDS/AML cells expressing ASXL1 and SETBP1 mutants (cSAM cells). After transduction of mouse genome-wide CRISPRa-v2 library, cSAM cells were cultured with DMSO or DAC for 1 week. We extracted genomic DNA from the cells at the start and end of our screen and quantified the relative representation of each sgRNA. To assess the effect of DAC on mitosis, we established the time-lapse system optimized for non-adherent MDS-L-2007 cells (a human secondary AML (sAML) cell line derived from an MDS cell line, MDS-92). Results: We first explored DepMap database and found that myeloid tumors are susceptible to DAC at clinically achievable concentration, while most solid tumor cell lines are not. The CRISPR/dCas9-activation screen revealed that genes involved in chromosome segregation, including Cdk1, Cdc20, Cdca8 and Dsn1, were significantly enriched in DAC-resistant cells, indicating that overexpression of mitotic regulators conferred resistance to DAC in cSAM cells. We also found that DAC strongly induced aneuploidy and polyploidy (>4n) in myeloid tumors, especially in those with TP53 mutations and sAML cells. Furthermore, the time-lapse imaging revealed that DAC treatment prolonged the duration of metaphase and increased multipolar mitosis and abscission failure in MDS-L-2007 cells (Figure). Thus, DAC perturbs mitosis and has cytotoxic effects at low concentrations in myeloid tumors. Next, we assessed the role of DNMT1 in the DAC-induced mitotic defects in myeloid tumors. To our surprise, the DAC-induced aneuploidy and apoptosis were markedly attenuated in DNMT1-depleted myeloid tumors. In contrast, overexpression of Dnmt1 increased sensitivity of myeloid tumors to DAC. These results indicate that DAC perturbs mitosis not through DNMT1 degradation and subsequent DNA demethylation, but through DNMT1 itself in myeloid tumors. In line with this, chromatin-bound DNMT1 protein was retained even with DAC, whereas DAC induced rapid degradation of DNMT1 in the nucleoplasm fraction. Importantly, the Dnmt1-C1229A mutant, which loses the ability to induce DNMT1-DNA covalent bond formation, did not enhance the DAC-driven mitotic defects. In addition, DAC treatment increased frequency of cohesion loss between sister chromatids in MDS-L-2007 cells, as evidenced by the absence of centromere connection of sister chromatids at metaphase. Moreover, the in vitroXenopus cell-free system which reproduces replication-dependent DNA methylation maintenance revealed that addition of 5-aza-dCTP strongly promoted chromatin binding of DNMT1, while it inhibited that of cohesion complex members (SMC3 and SCC1). Taken together, these results strongly indicate that the aberrant DNMT1 retention on chromatin induced by DAC leads to cohesion loss and mitotic defects in myeloid tumors. Conclusion: These findings challenge the current assumption that DAC inhibits leukemogenesis by inducing DNA hypomethylation. Rather, the anti-leukemia effect of DAC is likely mediated mainly by the non-epigenetic mechanism: mitotic perturbation through aberrant DNMT1-DNA covalent bonds and cohesion loss of sister chromatids. Our study also suggests that DAC will be most effective on tumors with impaired mitotic fidelity, such as AML and MDS with TP53 mutations. Figure 1View largeDownload PPTFigure 1View largeDownload PPT Close modal