Persistent binding of DNMT1 to chromatin causes replication dysfunction in TP53-mutated myeloid neoplasms

Abstract

Abstract Mutations in TP53 are the worst prognostic factor for myeloid malignancies, with a median survival of 6-9 months. Treatment with a hypomethylating agent (decitabine or azacitidine) is the standard of care for TP53-MNs, but responses are usually transient and durable remissions uncommon. We previously reported that decitabine induces replication stress in TP53-mutated myeloid neoplasms (TP53-MN), and that TP53-MN are selectively sensitive to the combination of decitabine and ATR inhibition, with synergistic killing of TP53-mutated AML cell lines and primary TP53-MN observed both in vitro and in vivo. We also found that hypomethylation was not sufficient to produce this sensitivity, as a reversible DNMT1 inhibitor (DNMT1i) did not recapitulate this synergistic killing. Here, we report our efforts to characterize molecular mechanisms by which loss of TP53 sensitizes cells to decitabine-induced DNA replicative stress so that these mechanisms can be further exploited therapeutically. Decitabine functions through incorporation into DNA and mimics a hemi-methylated CpG, which is recognized by DNMT1, the major maintenance DNA-methyltransferase. DNMT1 forms a covalent linkage to this abnormal base, resulting in DNA-protein adducts, DNMT1 degradation, and subsequent global hypomethylation. In contrast, the DNMT1i functions by blocking the catalytic site of DNMT1, inducing marked DNA hypomethylation without producing DNA-protein adducts. Although DNMT1i had reduced cytotoxicity compared to decitabine, it induced replication stress selectively in TP53-deficient AML cells as measured by single molecule DNA fiber assays. Both treatments also showed increased fork stalling indicated by fork asymmetry and increased single-stranded breaks by alkaline comet assays in TP53-deficient cells. Interestingly, neither treatment disrupted replication origin firing, as has been previously reported in TP53-mutated cells. We then generated doxycycline-inducible shDNMT1 knockdown lines in both the wildtype and TP53-deficient backgrounds, to allow us to observe the effect of the transient loss of DNMT1 protein. While loss of DNMT1 protein did produce significant hypomethylation, no replication stress was observed. Based on these observations, we hypothesized that inhibition of catalytic activity of DNMT1 leads to its retention at replication forks, resulting in replication stress. To test this hypothesis, we performed subcellular fractionation Western blots for DNMT1. We found that DNMT1 protein in untreated cells is almost entirely in the soluble nuclear fraction, but after treatment with DNMT1i, more than half of DNMT1 protein is chromatin bound. To further demonstrate this association, we generated wildtype or TP53-deficient AML cells in which an HA tag was knocked-in to the carboxy-terminus of DNMT1. We observed DNMT1-HA foci by immunofluorescence significantly increased after decitabine and DNMT1i treatment, especially in TP53-deficient cells. Collectively, these data suggest that DNMT1 retention at replication forks is a novel mechanism regulating DNA replication. Inhibition of DNMT1 methylation activity results in its abnormal retention at replication forks, which selectively induces replication stress in TP53 mutant AML cells, representing a potential therapeutic vulnerability for TP53-mutated myeloid neoplasms.