Abstract Objectives: Acute myeloid leukemia (AML) is mostly driven by specific mutations that affect transcription factors and epigenetic regulation. While targeted therapies are being developed for few AML subtypes, most patients still rely on systemic genotoxic treatments to induce complete remission and/or to bridge to bone marrow transplantation. Hence, deeper understanding of underlying disease and therapeutic mechanisms is urgently needed. In recent years, agents targeting the DNA damage response - such as PARP inhibitors (PARPi) already exploited in solid tumours, have emerged as promising therapeutic agents also in AML. PARPi alone exhibits high efficacy for a subset of AML, such as those driven by AML1-ETO and RARα fusions, but is rather ineffective forMixed-Lineage Leukemia gene-rearranged (MLLr) AML. A major function of these genotoxic agents is to interfere with DNA replication resulting in replication fork collapse and excessive DNA damage. However, cancer cells may adapt damage tolerance mechanisms, including fork reversal and repriming, to potentially bypass the genotoxic insults. We recently reported that, upon stimuli-induced proliferation, hematopoietic stem and progenitor cells (HSPCs) experience drastic changes in DNA replication dynamics and use specific fork protection mechanisms to promote their repopulation potential. Hence, we hypothesized that AML derived from deregulated hyperproliferating HSPCs may also adapt their DNA replication process, which may offer novel vulnerabilities with therapeutic potential. Methods: We investigated DNA replication dynamics and replication-associated DNA damage in established human and mouse AML models, that express different oncofusion proteins and reportedly display differential sensitivity to PARPi. To this aim, we used single-cell and single-molecule assays (e.g. comet assays, quantitative image-based cytometry, DNA fiber spreading) under baseline conditions and upon treatment with PARPi or standard-of-care chemotherapeutics. We also combined these functional assays with inactivation or overexpression of key replication fork plasticity factors, and with cell survival, apoptosis and differentiation assays, to evaluate how specific mechanisms of fork plasticity contribute to replication dynamics and therapy response in different AML subtypes. Results: We found that all human and mouse AML lines exhibit remarkably fast replication fork progression, compared to control HSPCs. However, AML lines previously reported as PARPi-sensitive – i.e. NB-4 (PML-RARα) and Kasumi-1 (AML1-ETO) – rapidly respond to PARPi by further accelerating fork progression, and eventually experiencing fork breakage and S-phase specific DNA damage signaling. Remarkably, inactivation of the specialized fork restart helicase RECQ1 suppresses these phenotypes, linking PARPi response to replication fork reversal and restart. Conversely, we found that PrimPol-mediated DNA synthesis – an alternative DNA damage tolerance mechanism that competes with fork remodelling – is predominant in MLL-rearranged leukemia, i.e. an aggressive AML subtype associated with poor prognosis. Notably, overexpression of Primpol in PARPi-sensitive AML is sufficient to prevent fork breakage and PARPi-sensitivity in these cells, while Primpol downregulation sensitizes MLL-rearranged AML lines (e.g. MOLM-13, THP-1), inducing fork collapse and replication-associated breaks. Importantly, similar observations were collected upon AML standard-of-care treatments (e.g. AraC), suggesting that mechanisms underlying replication fork plasticity play a central role in governing AML response to genotoxic treatments. Conclusions: These findings shed light on key mechanisms of the replication stress response in AML and underscore their potential for clinical translation. In light of our data, fork plasticity factors may represent novel therapeutic targets and predictive biomarkers for AML patient stratification. This evidence encourages the development of inhibitory compounds to target specific replication fork transactions and thereby improve therapeutic outcomes.
Fung et al. (Mon,) studied this question.