Targeting PERK Arm of Unfolded Protein Response Helps to Eliminate Therapy-Induced Residual Senescent-like Acute Myeloid Leukemia Cells

Introduction Overcoming therapy resistance following rapid eradication of bulk AML requires efficient elimination of persistent cells that survive initial treatment. Therapy-resistant AML cells represent minimal residual disease (MRD) and underlie emergence of relapse after varying periods of remission. Further, leukemia stem cells (LSCs) residing within MRD, with their inherent long-term self-renewal capacity, also account for AML relapse and drug resistance. Recent findings indicate that chemotherapy (Ara-C) induces senescence-like resilient cells capable of initiating AML recurrence (Duy et al., Cancer Discovery 2021). We have shown that these senescent cells highly upregulate BCL-2 and BCL-xL for survival and can be targeted by the dual BCL-2/xL degrader (Jia et al. Haematologica 2023). Given that leukemic cells can reduce their dependence on BCL-2 by upregulating other survival pathways, new strategies for targeting MRD cells are needed to increase overall survival of AML patients. The unfolded protein response (UPR) plays a key role in tumor adaptation to stress conditions like high rates of proliferation, increased demands for protein production and chemotherapy. UPR-mediated pro-survival program is uniquely wired in functioning and self-renewal of hematopoietic stem cells (HSCs). Activation of ATF4, a terminal effector of PERK-eiF2α branch of UPR allows HSCs but also LCSs to survive amino acid deprivation (van Galen et al., Cell Rep 2018). However, to date, the role of UPR in leukemia MRD has been poorly investigated. Here, our preliminary in vitro results demonstrate that targeting PERK pathway may eradicate post-therapy residual leukemia. Results First, we tracked proliferation and viability of MOLM-14, MV4-11 and MOLM-13 AML cells exposed to Ara-C (0.1 µM) using IncuCyte live-cell imaging. Long-term (up to 10 days) treatment yielded a population of non-proliferating viable cells. C12-FDG staining revealed that a subset of surviving cells (~15%) expressed senescence-associated β-galactosidase activity, which was accompanied by upregulation of cell cycle inhibitor p21, anti-apoptotic Bcl-xL and markers of senescence-associated secretory phenotype (IL-8, IL-1β, CCL5). FACS-sorted post Ara-C senescent-like MOLM-14 cells (C12-FDG-high fraction) showed increased mRNA level and phosphorylation of PERK, a hallmark of UPR activation, as well as upregulation of ATF4, a downstream component of PERK-eiF2α pathway. Interestingly, C12-FDG-high senescent cells also expressed high levels (mRNA and protein) of NRF2, a non-canonical down-stream PERK target, which is a key transcription factor regulating oxidative stress and inflammatory responses. Consistent with previous study (Duy et al. Cancer Discovery 2021), post Ara-C senescent cells recovered and re-expanded over the course of 10 days in medium without the drug. Importantly, shRNA-mediated knock-down of PERK significantly reduced viability of C12-FDG-high senescent cells in comparison to C12-FDG-low PERKKD and scramble shRNA controls. Next, we analyzed the relevance of UPR pathway for chemotherapy resistance in AML patient specimens. GSEA analysis of scRNA-seq of residual bone marrow AML cells isolated from patients after induction chemotherapy (GSE116256 Takao et al. BioRxiv 2024), indicated a significant enrichment in UPR gene signature in one of the two tested cases. Additionally, we found significant up-regulation of UPR-related genes in 2 out of 3 FACS-sorted non-proliferating quiescent leukemic stem cell populations (CFSE-label retaining cells) isolated from AML PDX models. Conclusions Our study shows that senescent-like AML cells arising after chemotherapy upregulate PERK canonical (via eiF2α/ATF4) and non-canonical (via NRF2) pathways, and depletion of PERK reduces their survival. We also suggest that targeting UPR may have the potential to eliminate not only residual AML but also LSCs associated with disease propagation. In vivo efficacy studies of PERK depletion in label-traced non-proliferating senescent-like AML and quiescent LSCs are ongoing and will be reported.