DNMT3A Mutations Deregulate the DNA Demethylation Pathway in Acute Myeloid Leukemia

Introduction: Mutation of DNMT3A, encoding a de novo methyltransferase essential for cytosine methylation, is a common early event in clonal hematopoiesis (CH) and adult acute myeloid leukemia (AML). Spontaneous deamination of methylated cytosines incurs DNA damage (methylation damage), which is repaired by the base excision repair enzymes MBD4 and TDG. Wildtype DNMT3A binds TDG, thereby potentiating its repair activity. In previous work (Massaar et al., EHA 2024), we had found that whereas wildtype DNMT3A stimulates TDG activity, mutant DNMT3A impairs TDG-mediated repair of methylation damage in vitro. Besides its role in the repair of methylation damage, TDG is also involved in another branch of the DNA demethylation pathway. In this branch, TET enzymes catalyze a stepwise hydroxylation process in which 5mC is converted to 5-hydroxymethylcytosin (5hmC). This is then further hydroxylated to form 5-formylcytosine (5fC), followed by 5-carboxylcytosine (5caC). Based on our previous findings, we aimed to further dissect the mechanisms by which mutant DNMT3A inhibits TDG activity. In addition, based on our initial results, we investigated whether the hydroxylation-demethylation pathway is similarly disrupted by mutation of DNMT3A. Finally, we hypothesized that in both branches of the demethylation pathway, the inhibiting effect of mutant DNMT3A on TDG was the result of decreased binding activity of TDG to its canonical substrates.

Methods: AlphaFold 2.0 was used to predict changes in the interacting residues between TDG and mutant DNMT3A. Glycosylase activity assays with 5caC as a substrate were performed to confirm the role of TDG in demethylation through hydroxylation. Band shift assays were performed to assess the binding affinity of TDG to oligoduplexes in the presence of wildtype or mutant DNMT3A, using either the G:T mismatch or 5caC as substrate, to represent the two branches of the demethylation pathway in which TDG acts.

Results: Alphafold 2.0 predicted that mutation of DNMT3A results in novel interactions with TDG. Interestingly, most forms of mutant DNMT3A are predicted to interact with residues located within the uracil-DNA glycosylase domain of TDG, whereas wildtype DNMT3A is predicted not to interact with this specific domain. Given the significance of this domain for TDG repair capacity, such changes may further impact TDG repair activity. Glycosylase activity assays including 5caC as a substrate confirmed that TDG acts in the hydroxylation-demethylation pathway. Using a band shift assay we observed that increasing concentrations of recombinant human wildtype DNMT3A stimulates TDG binding to oligoduplexes containing a G:T-mismatch. Of note, we confirmed a decreased binding affinity of TDG to the same oligoduplexes when co-titrated with mutant DNMT3A (R635W, R688C, R882C and A884V) compared to wildtype DNMT3A. This could potentially explain how mutant DNMT3A inhibits TDG repair activity. Since TDG is an integral part of the hydroxylation-demethylation pathway, we next investigated whether wildtype and mutant DNMT3A differentially impact TDG binding activity to oligoduplexes containing a 5caC as substrate. Strikingly, similar to our previous experiments, we observed that mutant DNMT3A negatively impacts TDG binding activity towards this pivotal epigenetic derivative. Finally, similar to previous work, we observe that the DNMT3A R635W mutation, not involved as residue in the TDG-DNMT3A interaction interface, only moderately impacts TDG binding activity, independent of its substrate. This indicates that the effect on TDG binding may depend on whether the mutated DNMT3A residue is involved in the TDG-DNMT3A interaction.

Conclusion & Discussion: We here provide insights in the potential mechanisms underlying TDG inhibition by mutant DNMT3A. Our results show that mutated DNMT3A causes alterations in the predicted interaction between DNMT3A and TDG, which may interfere with the glycosylase function of TDG. We also confirmed decreased binding affinity of TDG to its canonical substrates in the presence of mutant DNMT3A, representing a potential mechanism behind the observed inhibitory effect of mutant DNMT3A on TDG activity. Finally, we showed that mutant DNMT3A not only impairs methylation damage repair, but likely also active demethylation through hydroxylation, uncovering another potential mechanistic link between mutant DNMT3A and leukemogenesis.

Versluis:Novartis: Honoraria; Abbvie: Honoraria; Rigel: Membership on an entity's Board of Directors or advisory committees; ExcelThera: Membership on an entity's Board of Directors or advisory committees.