A genome-wide CRISPR screen identifies conventional and novel ferroptosis regulators in Acute Myeloid Leukemia Free

Background: Acute myeloid leukemia (AML) often acquires drug resistance and relapses after standard apoptosis-based therapy. Ferroptosis is an iron-dependent non-apoptotic form of cell death, with lipid peroxidation as a hallmark. Accumulating evidence has been proving the potential of ferroptosis for AML therapy that replenishes current therapeutic strategies and overcomes acquired resistance. Glutathione Peroxidase 4 (GPX4) is an antioxidant protein, reduces lipid peroxidation in cell membranes and functions as the major negative regulator of ferroptosis. Our data has shown that the inhibition of GPX4 with a specific inhibitor, ML210, induces mitochondria-dependent ferroptosis even in venetoclax-resistant AML cells. However, the prolonged survival of AML xenograft mice by GPX4 knockdown in vivo is limited. These findings underscore the need to further optimize ferroptosis-based approaches to improve anti-leukemic efficacy.

Methods: To identify novel therapeutic targets that may enhance ferroptosis-based AML treatment, we performed a genome-wide CRISPR knockout screen in MOLM13 cells in the context of ferroptosis induction to identify AML-specific ferroptosis regulators. To minimize compound off-target and dosage effects, in this study we induced ferroptosis by genetic GPX4 knockdown with a previously verified doxycycline-inducible short-hairpin RNA (shRNA). We engineered MOLM13 cells with stable Cas9 expression, and the verified cell line was transduced with the lentiviral Brunello library targeting 19,114 genes. After 4 days of puromycin selection, the cells were treated with 1 µg/ml doxycycline for GPX4 knockdown. Cells were passaged every 3 days with the single guide RNA (sgRNA) coverage in each passage at about 500-fold. Ferroptotic cell death was monitored every 3 days during passaging. Genomic DNA was prepared from the cells collected after 9 days of doxycycline treatment and used as the template for amplifying the sgRNA library with a modified two-round of PCR. After high-throughput sequencing, the sgRNA enrichment was analyzed by MAGeCK with comparison to the non-doxycycline control.

Results: Our results successfully identified several known pro-ferroptosis regulators, such as ACSL4, DHCR7, and LPCAT3 from positive selection. We also observed multiple established ferroptosis suppressors, such as SLC7A11 and GPX4-independent ferroptosis suppressors, such as DHODH, AIFM2, and GCH1 from the negative selection. Interestingly, we identified the ferroptosis suppressors, LRP8 and STARD7, which have been recently investigated in solid tumors, but not in AML. These results indicate the robustness of our screening dataset. Interestingly, OPA1 is identified as a potential ferroptosis suppressor in the present AML study, opposite to the pro-ferroptosis role reported in mouse embryonic fibroblasts and human U2OS osteosarcoma cells, suggesting the specific role of OPA1 in AML. Further enrichment analysis showed that multiple mitochondrion-related pathways were top-ranked, including mitochondrial RNA/protein degradation, mitochondrial gene expression, mitochondrion organization and complex I biogenesis. Notably, we found a novel potential ferroptosis suppressor, MAFG, a basic leucine zipper transcription factor, ranked as a top hit. Based on re-analyses of publicly available databases, the expression of MAFG is positively correlated with GPX4 expression in AML and with the resistance to several ferroptosis inducers, including ML162, ML210, and RSL3. ChIP-seq data of MAFG in K562 cells indicates SLC7A11 is one of its targets. Therefore, we hypothesize that MAFG is the upstream transcriptional regulator of SLC7A11, modulating ferroptosis in a GPX4-dependent manner.

Conclusion: We performed a genome-wide CRISPR knockout screen using a GPX4 genetic knockdown model and identified multiple known and novel ferroptosis regulators in AML cells. Specifically, our data suggests MAFG as a novel potential ferroptosis suppressor in AML. Further mechanistic studies are ongoing.