DNA Cytosine-Demethylating Agent 5-Aza-2'-Deoxycytidine Targets Leukemia Cells through Reducing DNA N⁶-Methyladenine

Introduction: It is known that overexpression of DNA methyltransferases (e.g., DNMT1) is frequent and changes of DNA cytosine methylation (5mC) are a constant feature of cancers. DNA methylation inhibitors, such as 5-aza-2'-deoxycytidine (Dec), have been in clinics for patients with leukemia. It is classically believed that promoter hypomethylation coupled by reexpression of epigenetically-suppressed tumor suppressors is a core mechanism behind Dec-impaired leukemia cell growth. However, the fact that global DNA methylation profiling barely predicts Dec-response suggests a demethylation-independent mechanism of Dec-induced cell death. N6-methyladenine (m6A) has been identified recently as an abundant DNA modification in eukaryotes (Wu, Nature 2016;532:329). Importantly, m⁶A is extensively present in the human genome, and m⁶A abundance is associated with tumorigenesis (Xie, Cell 2018;71:306). Furthermore, the DNA m⁶A is dynamically modulated by the methyltransferases (i.e., METTL3, N6AMT1) and demethylases (i.e., ALKBH1), and changes in m⁶A predict gene expression (Wu, Nature 2016;532:329). Given a potential crosstalk between m⁶A and distinct epigenetic mechanisms (Yao, Nat. Commun 2017;8:1122), we hypothesized that the anticancer actions of Dec may partially result from changes in DNA m⁶A in leukemia cells.

Methods: Protein expression of target genes was assessed by Western blotting. The levels of DNA cytosine methylation (5mC) and N⁶-methyladenine (m⁶A) were measured by dotblotting or liquid chromatography-mass spectrometry (LC-MS/MS). The cell viability and apoptosis were determined by the Cell Counting Kit 8 (CCK8) assays as well as the Annexin V/Propidium Iodide staining and flow cytometry. The peripheral blood mononuclear cells (PBMCs) of leukemia patients from Mayo Clinic were prepared by Ficoll-Hypaque gradient centrifugation.

Results: To test our hypothesis, leukemia cells, Kasumi-1, MV4-11, K562 and KU812, were exposed to 2 µM Dec, a clinical achievable concentration, for 72 hours. As expected, Dec treatment led to a downregulation of DNMT1 and DNMT3a, a reduction of 5mC levels by dotblotting using anti-5mC antibody, a blockage of cell proliferation and a promotion of cell apoptosis. When genomic DNA was subjected to dotblotting using anti-m⁶A antibody, the results revealed a marked decrease of DNA m⁶A levels in all Dec-treated cells. Then genomic DNA from K562 and MV4-11 cells was enzymatically digested to 2'-deoxynucleosides. dA was quantified by HPLC-UV, while the amount of m⁶A was measured by isotope dilution HPLC-ESI-MS/MS using 15N labeled internal standard. The standard curves were generated using pure standards, from which the m⁶A/A ratio was calculated. In agreement with dotblotting results, Dec treatment significantly decreased DNA m⁶A abundance in both cell lines. Mechanistically, exposure to Dec led to a consistent increase of demethylase fat mass and obesity-associated protein (FTO), but not METTL3 nor ALKBH1 and ALKBH5. Further, knockdown of FTO increased DNA m⁶A, which was further confirmed by treatment with FTO inhibitors rhein and meclofenamic acid (MA). These data indicate that FTO may be responsible for Dec-induced m⁶A changes in leukemia cells.

To investigate the clinical implications of DNA m⁶A, we obtained PBMCs from AML patients (n = 10), who received Dec therapy (20 mg/m² daily for 5 days every 4 weeks) in Mayo Clinic. These PBMCs were further cultured for 48 hours, frozen and stored in 100% ethanol before DNA extraction. The results from dotblotting using anti-5mC or anti-m⁶A showed that a trend of decrease in both m⁶A and 5mC abundance is observed, and the pattern of changes in m⁶A and 5mC displays a positive correlation. Finally, exposure of leukemia cells to the combination of Dec (2 µM) with FTO inhibitor MA (50 µM) induced more cell apoptosis and greater inhibition on cell proliferation as compared to single agent in vitro, supporting FTO inhibitors as new therapeutic agents in leukemia.

Conclusion: Our studies suggest that the FTO-DNA m⁶A axis may partially mediate the therapeutic outcomes of Dec in leukemia. Our findings provide a new mechanistic paradigm for the anticancer activities of Dec, and define the m⁶A methylation status in leukemia cells as a new pharmacodynamic marker for their response to Dec-based therapy, pointing to a novel treatment strategy for incorporating m⁶A modulators to enhance the therapeutic index of Dec.