Abstract Background: Doxorubicin (DOXO) efficacy in triple-negative breast cancer (TNBC) is severely limited by acquired resistance and dose-limiting cardiotoxicity. We identified ceramide synthase 2 (CerS2), producer of tumor-suppressive very long-chain ceramides (VLCCs), as a prognostic marker and therapeutic target in breast cancer (BC). We previously developed DH20931, a CerS2 activator that inhibits growth across BC subtypes by inducing lipotoxic endoplasmic reticulum (ER) stress. We hypothesized that DH20931 could sensitize BC cells to DOXO and reverse resistance. Methods: The synergistic efficacy of DH20931 and DOXO was evaluated in parental and DOXO-resistant BC cell lines, e.g., MDA-MB-231, 4T1. Synergy was quantified using the Chou-Talalay method (Combination Index, CI). Mechanisms were assessed via confocal microscopy (DOXO uptake), immunoblotting (ER stress, apoptosis markers) and in vivo efficacy. Results: The DH20931 combination significantly reduced the required IC50 of DOXO by 3-8-fold. DH20931 demonstrated strong synergy with DOXO (CI <1) across all tested BC cell lines, including DOXO-resistant models. The synergy was eliminated in CerS2-knockout cells indicating CerS2 target engagement. Mechanistically, the combination therapy resulted in hyper-activation of the UPR, maximized ATF4/CHOP/PUMA expression, and a marked increase in cleaved caspase-3 compared to either agent alone. In vivo, the DH20931+DOXO combination significantly inhibited tumor growth compared to DH20931 and DOXO monotherapy. Conclusion: DH20931 effectively sensitizes BC cells to Doxorubicin and reverses DOXO-resistance, likely by enhancing CerS2-mediated lipotoxic ER-stress. This synergistic approach offers a promising clinical strategy to improve DOXO efficacy, reduce its cardiotoxicity by allowing dose reduction, and improve outcomes for patients with TNBC and other BC subtypes. Citation Format: Hissah Alatawi, Haritha H. Nair, Lingbao Ai, Abhisheak Sharma, Christopher Vulpe, Arun K. Sharma, Coy D. Heldermon, Satya Narayan. A novel CerS2 activator, DH20931, synergizes with doxorubicin to overcome therapeutic resistance in breast cancer [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2026; Part 1 (Regular Abstracts); 2026 Apr 17-22; San Diego, CA. Philadelphia (PA): AACR; Cancer Res 2026;86(7 Suppl):Abstract nr 4457.

Triple-negative breast cancer (TNBC) poses a significant therapeutic challenge because of the lack of defined molecular targets. Although ceramide synthase 2 (CerS2) plays a complex role in oncology, enhancing its enzymatic activity to produce proapoptotic, very long-chain ceramides (VLCC) is a potential anticancer strategy. In this study, we identify and characterize DH20931, a novel biisoquinoline derivative, as a newly identified small-molecule activator of CerS2. We provide genetic and biochemical evidence that CerS2 is the direct target of DH20931, which shows an effective, receptor-independent cytotoxicity across diverse breast cancer cell lines while sparing normal cells. In vivo, DH20931 demonstrates consistent tumor growth inhibition in both orthotopic xenograft and clinically relevant patient-derived xenograft models of TNBC, supported by a favorable safety and pharmacokinetic profile. Mechanistically, DH20931 triggers an effective dual mechanism of apoptosis. First, the accumulation of VLCCs induces lipotoxic endoplasmic reticulum (ER) stress, activating the proapoptotic ATF4-CHOP pathway. Second, we uncovered a previously unknown physical interaction between CerS2 and the ER calcium channel IP3R1. DH20931 promotes this interaction, enhancing ER-mitochondria proximity and facilitating a marked flux of Ca2+ into the mitochondria, which serves as an effective, secondary apoptotic signal. These findings validate CerS2 as a bona fide druggable target and present DH20931 as a promising clinical candidate. This unique synergistic mechanism, coupling lipotoxicity with calcium dysregulation, offers a convincing new strategy for treating aggressive and therapy-resistant breast cancers.

CRISPR/Cas9 screening is a powerful tool used to delineate genes critical for AML progression and drug resistance. However, most of the studies have been performed in cell lines and PDX models. While cell lines fail to represent the genetic, epigenetic, and phenotypic heterogeneity of primary leukemic cells, PDX models are limited by variable engraftment efficiency, time-consuming processes, and high costs. To overcome these challenges, we pioneered an integrated approach combining bioinformatics and CRISPR/Cas9 screening to uncover genetic drivers and vulnerabilities in AML using patient-derived primary cells in culture. Insights from our study could inform the development of targeted therapies and improve patient outcomes. Custom CRISPR/Cas9 screen targeting 2,440 genes associated with AML, leukemic stemness, and the druggable genome was performed using mononuclear cells (MNCs) isolated from peripheral blood or bone marrow from 9 AML patients. Briefly, the cells were transduced with custom library and cultured for 24-30 days. After an initial few days in suspension, a subset of MNCs from 5 of 9 patients transitioned to an adherent state 5-7 days post-transduction. Both adherent and suspension cells were collected at different time points, sgRNA was quantified and sequenced using the Illumina NovaSeq X Plus. sgRNA abundance was analyzed using the RRA algorithm of the MAGeCK computational tool and genes with significant negative or positive enrichment in each patient was identified by comparing early versus late time points and adherent versus suspension cell fractions. In a subset of patient samples, AML cells transitioned from suspension to an adherent phenotype during long-term culture, a state previously associated with increased chemotherapy resistance due to interactions with endothelial niches. Our CRISPR/Cas9 screen revealed that this adherent cell population harbored distinct genetic dependencies, including consistent negative enrichment of EDNRA and SOD1.These genes were selectively essential in the adherent fraction but not in suspension cells, suggesting context-specific vulnerabilities. The identification of these targets offers new opportunities to therapeutically disrupt chemo resistant AML niches, potentially enhancing the efficacy of existing treatments and reducing relapse risk. EDNRA has been implicated in many cancers and in context of AML, it is a downstream target of HOXA9 and MEIS1. Our results indicate that EDNRA knockout inhibits AML cell proliferation and growth, highlighting its potential as a therapeutic target. Furthermore, EDNRA is overexpressed in AML compared to normal cells, and since FDA approved EDNRA inhibitors such as Macitentan, Bosentan, Ambrisentan, and Clazosentan are already being used to treat pulmonary arterial hypertension, this presents a promising opportunity to evaluate these inhibitors as potential chemotherapy-sensitizing agents for high-risk AML.SOD1 plays a critical role in eliminating toxic radicals generated within biological systems and has been associated with poor outcomes in AML. SOD1 has been identified as a synthetic lethal target in PPM1D-mutant leukemia cells, corroborating our findings. TRAPPC9, SUZ12, and ERG were significant genes with negative enrichment when comparing adherent and suspension fractions across 5 patient samples. These genes have a higher expression in primary AML cells compared to normal controls, implying potential roles in disease progression. SDHA was identified as a significant negatively enriched gene in the suspension cell fraction in 4 of 9 patients. It exhibited elevated expression in AML cells compared to controls, and its deficiency has been linked to the development of drug resistance in AML, emphasizing its role in both drug resistance and disease progression. This study demonstrates the establishment of a scalable platform for CRISPR screening in primary AML cells for identification of more conserved vulnerabilities that may be exploited therapeutically, with EDNRA as a potential target. Ongoing work includes validating key targets with the goal of advancing therapeutic strategies. Future studies will expand CRISPR screening to a broader set of patient samples with diverse cytogenetic profiles and incorporate CROP-seq technology to perform single-cell CRISPR screens, providing deeper insights into the molecular mechanisms underlying AML progression and relapse.

Abstract Cytarabine, daunorubicin, and etoposide (ADE) have been the standard backbone of induction chemotherapy regimen for patients with pediatric acute myeloid leukemia (pAML) for >5 decades. However, chemoresistance is still a major concern, and a significant proportion of pAML becomes resistant to ADE treatment and relapse, leading to poor survival. Therefore, there is a considerable need to identify mechanisms mediating drug resistance for overcoming chemoresistance. Herein, we performed synthetic lethal CRISPR/Cas9 screens using the ADE components to identify response markers. We further integrated significant markers in 3 independent pAML clinical cohorts treated with only an ADE regimen to identify drug response biomarkers with prognostic significance. We were able to identify several mediators that represent clinically and biologically significant marker genes for ADE treatment, such as BCL2, CLIP2, and VAV3, which are resistant markers to ADE, with high expression associated with poor outcomes in pAML treated with ADE, and GRPEL1, HCFC1, and TAF10, which are sensitive markers to ADE, with high expression showing beneficial outcomes. Notably, BCL2, CLIP2, and VAV3 knockdowns in their expression in AML cell lines sensitized the cells more to the ADE components, suggesting that these modulators should be further studied as potential therapeutic targets to overcome chemoresistance.

Acute myeloid leukemia (AML) is a genetically diverse and aggressive blood cancer. FLT3 mutations occur in ~30% of AML cases and are targeted by tyrosine kinase inhibitors (TKIs) such as midostaurin, sorafenib, and gilteritinib across treatment stages. Gilteritinib, midostaurin, and sorafenib are FLT3-targeting tyrosine kinase inhibitors (TKIs); gilteritinib and midostaurin are type I TKIs active against both FLT3-ITD and FLT3-TKD, while sorafenib is a type II TKI with broader multi-kinase activity. Midostarurin is added to the standard care of chemotherapy and transplant, while gilteritinib is used in relapsed/refractory AML, and sorafenib in frontline or maintenance settings. Despite these advances, resistance remains a significant challenge due to secondary mutations and alternative pathway activation. In this study, we leveraged CRISPR/Cas9 screening of FLT3-ITD+ve AML cells to identify molecular pathways underlying resistance to FLT3-TKIs. To identify FLT3 inhibitor response modulators, we performed a custom CRISPR/Cas9 screen in FLT3-ITD⁺ AML cell lines MOLM-13 and MV4-11 using a library targeting 2,440 AML-relevant and druggable genes. Following puromycin selection, cells were exposed to sublethal FLT3 inhibitors (sorafenib, midostaurin, or gilteritinib) for approximately seven cell doublings. Genomic DNA was PCR-amplified and sequenced at 500× coverage. MAGeCK-RRA was used to compute gene-level RRA scores using the ReadRRA() function, defined as –log₁₀ of the more significant p-value from either positive or negative selection. Genes were then classified as resistant or sensitizing based on sgRNA depletion or enrichment, respectively, using a ±1.5 SD cutoff from the mean RRA score distribution. Our screen revealed a core set of modulators that reproducibly influence FLT3 inhibitor response regardless of drug or cellular context. Overall, 28 common resistance and 26 common sensitive genes were associated across the two cell lines and the three FLT3-TKIs tested. Of the 28 resistant genes, 6 are mapped to drugs approved or in trials, highlighting the potential therapeutic approaches to enhance the efficacy of TKIs. These include CDK6 (cyclin-dependent kinase 6), MDM2 (an oncogene), and TP53 with roles in multiple pathways, including oncogene-induced senescence, cell cycle, DNA methylation, and transcriptional repression, etc. Targeting these pathways and modulating p53 and microRNA regulation, which were also enriched in our pathway analysis, may help overcome resistance to FLT3 inhibitors in AML. FLT3 inhibitors induce resistance via p53 destabilization through the STAT5–MDM2 axis; consistent with current results, thus, the combination with MDM2 inhibitors has the potential to overcome resistance to FLT3-TKIs. DHODH, a dehydrogenase, and FASN, fatty acid synthase, play a role in metabolic adaptation to FLT3 inhibition, and drugs targeting these can potentially improve the therapeutic effects of FLT3 inhibitors. NPEPPS, an aminopeptidase, and TBCE emerged as druggable dependencies involved in chemoresistance and may serve as therapeutic targets to enhance treatment efficacy in AML. Conversely, MTIF2 and PEX16, involved in mitochondrial and peroxisomal metabolism, may function as biomarkers of FLT3 inhibitor resistance and merit further validation. In parallel, ASXL2, PTEN, and NF1 were identified as consistent sensitizer genes in our screen, consistent with their known tumor suppressor roles. These findings highlight key vulnerabilities and resistance mechanisms in FLT3-mutant AML and suggest actionable pathways and biomarkers for improving targeted therapy outcomes. CRISPR/Cas9 screen identified effective genetic modulators of FLT3 inhibitor response in acute myeloid leukemia (AML). These findings reveal critical mechanisms underlying resistance, laying the groundwork for rational combination therapies to enhance treatment outcomes in FLT3-mutated AML. Future studies will validate these targets using gene knockdown in resistant and sensitive models to evaluate their impact on FLT3 inhibitor response. .

Physiologically relevant in vitro models are a priority in predictive toxicology to replace and/or reduce animal experiments. The compromised toxicant metabolism of many immortalized human liver cell lines grown as monolayers as compared to in vivo metabolism limits their physiological relevance. However, recent efforts to culture liver cells in a 3D environment, such as spheroids, to better mimic the in vivo conditions, may enhance the toxicant metabolism of human liver cell lines. In this study, we characterized the dynamic changes in the transcriptome of HepG2/C3A hepatocarcinoma cell spheroids maintained in a clinostat system (CelVivo) to gain insight into the metabolic capacity of this model as a function of spheroid size and culture time. We assessed morphological changes (size, necrotic core), cell health, and proliferation rate from initial spheroid seeding to 35 days of continuous culture in conjunction with a time-course (0, 3, 7, 10, 14, 21, 28 days) of the transcriptome (TempO-Seq, BioSpyder). The phenotypic characteristics of HepG2/C3A growing in spheroids were comparable to monolayer growth until similar to Day 12 (Day 10-14) when a significant decrease in cell doubling rate was noted which was concurrent with down-regulation of cell proliferation and cell cycle pathways over this time period. Principal component analysis of the transcriptome data suggests that the Day 3, 7, and 10 spheroids are pronouncedly different from the Day 14, 21, and 28 spheroids in support of a biological transition time point during the long-term 3D spheroid cultures. The expression of genes encoding cellular components involved in toxicant metabolism and transport rapidly increased during the early time points of spheroids to peak at Day 7 or Day 10 as compared to monolayer cultures with a gradual decrease in expression with further culture, suggesting the most metabolically responsive time window for exposure studies. Overall, we provide baseline information on the cellular and molecular characterization, with a particular focus on toxicant metabolic capacity dynamics and cell growth, of HepG2/C3A 3D spheroid cultures over time.

PURPOSE:Bone is susceptible to fluctuations in iron homeostasis, as both iron deficiency and overload are linked to poor bone strength in humans. In mice, however, inconsistent results have been reported, likely due to different diet setups or genetic backgrounds. Here, we assessed the effect of different high and low iron diets on bone in six inbred mouse strains (C57BL/6J, A/J, BALB/cJ, AKR/J, C3H/HeJ, and DBA/2J). METHODS:Mice received a high (20,000 ppm) or low-iron diet (∼10 ppm) after weaning for 6-8 weeks. For C57BL/6J males, we used two dietary setups with similar amounts of iron, yet different nutritional compositions that were either richer ("TUD study") or poorer ("UCLA study") in minerals and vitamins. After sacrifice, liver, blood and bone parameters as well as bone turnover markers in the serum were analyzed. RESULTS:Almost all mice on the UCLA study high iron diet had a significant decrease of cortical and trabecular bone mass accompanied by high bone resorption. Iron deficiency did not change bone microarchitecture or turnover in C57BL/6J, A/J, and DBA/2J mice, but increased trabecular bone mass in BALB/cJ, C3H/HeJ and AKR/J mice. In contrast to the UCLA study, male C57BL/6J mice in the TUD study did not display any changes in trabecular bone mass or turnover on high or low iron diet. However, cortical bone parameters were also decreased in TUD mice on the high iron diet. CONCLUSION:Thus, these data show that cortical bone is more susceptible to iron overload than trabecular bone and highlight the importance of a nutrient-rich diet to potentially mitigate the negative effects of iron overload on bone.

Mechanistic toxicology is key to understanding chemical exposure-associated risks. However, mechanisms of toxicity remain unclear for many ubiquitous chemicals. Genetic diversity adds another level of complexity due to alterations in toxicant susceptibility. The emergence of the state-of-the-art CRISPR technologies has revolutionized toxicological research and facilitated delineation of complex chemical-gene interaction networks. A wide range of targeted in vitro CRISPR-based applications can be utilized in toxicological studies and enable elucidation of mechanisms of action, identification of susceptibility modulators and establishment of novel methodologies for toxicity testing. In addition, genome-wide pooled CRISPR screens allow comprehensive interrogation of genes for functions relevant to cytotoxic responses. Further, the adaption of CRISPR-based gene editing tools to in-vivo model organisms considerably improve the translational potential of functional toxicogenomics. Finally, integration of endpoints generated by multiple CRISPR-enabled toxicological studies will constitute a critical component of risk assessment that can support extrapolation to human hazard and determination of adverse outcomes.

Titanium dioxide nanoparticles (nTiO2) have been considered a possible carcinogen to humans, but most existing studies have overlooked the role of human enzymes in assessing the genotoxicity of nTiO2. Here, a toxicogenomics-based in vitro genotoxicity assay using a GFP-fused yeast reporter library was employed to elucidate the genotoxic potential and mechanisms of nTiO2. Moreover, two new GFP-fused yeast reporter libraries containing either human CYP1A1 or CYP1A2 genes were constructed by transformation to investigate the potential modulation of nTiO2 genotoxicity in the presence of human CYP enzymes. This study found a lack of appreciable nTiO2 genotoxicity as indicated by the yeast reporter library in the absence of CYP expression but a significantly elevated indication of genotoxicity in either CYP1A1- or CYP1A2-expressing yeast. The intracellular reactive oxygen species (ROS) measurement indicated significantly higher ROS in yeast expressing either enzyme. The detected mitochondrial DNA damage suggested mitochondria as one of the target sites for oxidative damage by nTiO2 in the presence of either one of the CYP enzymes. The results thus indicated that the genotoxicity of nTiO2 was enhanced by human CYP1A1 or CYP1A2 enzyme and was associated with elevated oxidative stress, which suggested that the similar mechanisms could occur in human cells.

Past functional toxicogenomic studies have indicated that genes relevant to membrane lipid synthesis are important for tolerance to the lanthanides. Moreover, previously reported imaging of patient's brains following administration of gadolinium-based contrast agents shows gadolinium lining the vessels of the brain. Taken together, these findings suggest the disruption of cytoplasmic membrane integrity as a mechanism by which lanthanides induce cytotoxicity. In the presented work we used scanning transmission electron microscopy and spatially resolved elemental spectroscopy to image the morphology and composition of gadolinium, europium, and samarium precipitates that formed on the outside of yeast cell membranes. In no sample did we find that the lanthanide contaminant had crossed the cell membrane, even in experiments using yeast mutants with disrupted genes for sphingolipid synthesis-the primary lipids found in yeast cytoplasmic membranes. Rather, we have evidence that lanthanides are co-located with phosphorus outside the yeast cells. These results lead us to hypothesize that the lanthanides scavenge or otherwise form complexes with phosphorus from the sphingophospholipid head groups in the cellular membrane, thereby compromising the structure or function of the membrane, and gaining the ability to disrupt membrane function without entering the cell. Graphical Abstract High-resolution electron microscope images and corresponding elemental maps showing substantial evidence of extracellular toxicity of lanthanides to exposed yeast cells.

In the last two decades, the development of nanotechnology has resulted in inorganic nanoparticles playing crucial roles in key industries, ranging from healthcare to energy technologies. For instance, gold and silver nanoparticles are widely used in rapid COVID-19 and flu tests, titania and zinc oxide nanoparticles are commonly found in cosmetic products, and superparamagnetic iron oxide nanoparticles have been clinically exploited as contrast agents and anti-anemia medicines. As a result, human exposure to nanomaterials is continuously increasing, raising concerns about their potential adverse health effects. Historically, the study of nanoparticle toxicity has largely relied on macroscopic observations obtained in different in vitro and in vivo models, resulting in readouts such as median lethal dose, biodistribution profile, and/or histopathological assessment. In recent years, omics methodologies, including transcriptomics, epigenomics, proteomics, metabolomics, and lipidomics, are increasingly used to characterize the biological interactions of nanomaterials, providing a better and broader understanding of their impact and mechanisms of toxicity. These approaches have been able to identify important genes and gene products that mediate toxicological effects, as well as endogenous functions and pathways dysregulated by nanoparticles. Omics methods improve our understanding of nanoparticle biology, and unravel mechanistic insights into nanomedicine-based therapies. This review aims to provide a deeper understanding and new perspectives of omics approaches to characterize the toxicity and biological interactions of inorganic nanoparticles, and improve the safety of nanoparticle applications.

CPX-351 (dual-drug liposomal encapsulation of daunorubicin and cytarabine in a synergistic 1:5 molar ratio) is approved for newly diagnosed high-risk and secondary AML as an alternative to the widely used cytarabine and daunorubicin (7+3) regimen. Despite newly approved drugs, many AML patients still developed treatment resistance and relapse, leading to poor survival. Amid mounting research into the molecular mechanisms of AML therapies, treatment resistance remains a major concern. To overcome treatment resistance, it is necessary to examine the molecular mechanisms underlying the CPX-351 and 7+3 treatment response. We conducted custom CRISPR/Cas9 synthetic-lethal screens targeting 2440 genes (AML-relevant genes, cytarabine and daunorubicin pharmacology, and druggable genes) in six AML cell lines, profiled drug-sensitive/-resistant modulators, and identified novel targets for overcoming drug resistance in CPX-351 and 7+3 regimen. Six AML cell lines were transduced with a custom library and puromycin was selected before pool transduced cells were treated with DMSO, the IC30-IC50 of cytarabine, daunorubicin, or CPX-351 for 7 doubling-times at 500x coverage. All samples were processed with genomic DNA, sgRNAs were amplified by PCR, and Illumina NovaSeq 6000 SP 100SR sequencing was performed. The abundance of sgRNA was analyzed using MAGeCK-RRA to estimate the RRA drug response score (Drug vs. Control conditions at 7-doubling times) or RRA essential score (Control at 7-doubling time vs. Control at Day 0). Genes with an average RRA score of < -1 across six AML cell lines were defined as drug-resistant or essential genes (negative selection), whereas genes with an average RRA score of > 1 were categorized as sensitive or suppressor genes (positive selection). Next, comparisons of functional pathways with over-representation analysis were performed on resistant and sensitive genes across three drugs to determine differences in their functions. At the average RRA response score cut-off, 61, 163, and 112 were identified as drug-resistant genes, while 46, 93, and 86 were sensitive genes to cytarabine, daunorubicin, or CPX-351 (provided by Jazz Pharmaceuticals), respectively. Fig 1A showed a Venn diagram of all significant resistant and sensitive genes across all three drugs with 14 common significant genes and 73, 174, and 118 unique genes to cytarabine, daunorubicin and CPX-351, respectively, along with their essential score. Among these significant common response genes, some notable genes included ABCC1, SAMHD1 and TP53, where CRISPR knocking out screens led to depletion or enrichment of sgRNA when treated with drugs across all AML cell lines. ABCC1 is a drug-resistant gene that encodes the ATP Binding Cassette C1-mediated drug efflux transporter; therefore, knocking out the efflux transporter causing drug resistance, would increase the lethality of AML cells . Our result also showed ABCC1 isa tumor suppressor gene; because knocking out the ABCC1 gene improves cell survival and growth . SAMHD1 is another notable gene that encodes for deoxynucleoside triphosphate (dNTP) tri-phosphohydrolase that cleaves physiological dNTPs into deoxyribonucleosides and inorganic triphosphate. SAMHD1 has been shown to cause resistance in cytarabine, but there is limited evidence of its response to other drugs. In this study, we showed that SAMHD1 is a resistant gene for CPX-351 and a sensitive gene for daunorubicin, and it is a tumor suppressor gene. Lastly, TP53 is one of the significant common sensitive genes to cytarabine, daunorubicin, and CPX-351. TP53 encodes for the well-known tumor suppressor gene tumor protein 53, and mutations in TP53 have been shown to affect drug response. In this study, we demonstrated that knocking out TP53 in all AML cell lines contributed to cell survival and growth as a tumor suppressor gene and a drug-sensitive gene across the three drugs. Comparing significant resistant and sensitive genes across CPX-351, cytarabine, and daunorubicin, as shown in Fig 1B, revealed that resistant genes shared some common involvement in the cell cycle and cellular response to stress, but resistant genes for each drug displayed more distinct functional pathways. Moreover, each drug's sensitive genes revealed distinct functional enrichment pathways. Our current research is focused on identifying novel drug targets to combat drug resistance in the CPX-351 and 7+3 regimens.

Cytarabine, daunorubicin, and etoposide (ADE) remain the standard backbone chemotherapy for pediatric AML. In addition to ADE, gemtuzumab ozogamicin (GO) and bortezomib have been evaluated in the COG's AAML0531 and AAML1031 clinical trials, respectively. Nonetheless, when these experimental agents were combined with ADE, ADE+GO had shown to improve EFS but not OS, while ADE+bortezomib displayed no additional benefit over ADE alone. Thus, treatment options for pediatric AML are still limited, and development of resistance is a major concern, resulting in poor clinical outcomes. To overcome treatment resistance, it is crucial to understand the underlying molecular mechanisms of ADE±GO or bortezomib. Herein, we conducted custom CRISPR/Cas9 synthetic-lethal screens targeting 2440 genes (including AML-relevant genes, pharmacologically relevant genes, and druggable genes), in six AML cell lines with exposure to ara-C, daunorubicin, etoposide, GO and bortezomib followed by integration of the results with outcome data from multiple clinical trials where induction treatment was either standard ADE chemotherapy, ADE+GO or ADE+bortezomib. Six AML cell lines were transduced with a custom library and puromycin selected. Then pool transduced cell populations were then treated with DMSO, the IC30-IC50 levels of cytarabine, daunorubicin, etoposide, GO, or bortezomib for 7-doubling-times at 500x coverage. All samples were processed with genomic DNA extraction, sgRNAs were amplified by PCR, and were subjected to Illumina NovaSeq 6000 SP 100SR sequencing. The abundance of sgRNA was analyzed using MAGeCK-RRA to estimate the RRA drug response score (drug vs. control conditions at 7-doubling times). Genes with an average RRA score of < -1 across six AML cell lines were defined as drug-resistant genes (negative selection), whereas genes with an average RRA score of > 1 were categorized as sensitive genes (positive selection). Next, we investigated leukemic cell gene expression levels of the drug-resistant and drug-sensitive genes with clinical outcomes. At the average RRA score cutoff from CRISPR screens, 459 genes met the significance as resistant or sensitive genes to at least one drug from ADE's screens as ADE genes, 244 genes were significant from ADE genes and GO's screen as ADE+GO genes, and 166 genes were significant from ADE genes and bortezomib'screen as ADE+bortezomib genes ( Fig 1a). ADE genes were evaluated with clinical outcomes with patients treated with ADE AML02 (n=163), ADE arm of AAML0531 (n=201) and AAML1031 (n=411); while ADE+GO genes were evaluated in ADE+GO arm of AAML0531 (n=205) and ADE+bortezomib genes were evaluated in ADE+bortezomib arm of AAML1031 (n=436). Genes showing drug resistance in CRISPR screens and associated with detrimental outcomes (EFS, OS, or MRD1) at high expression were deemed clinically and biologically important drug genes and vice versa. Our preliminary results showed 21 met the criteria as clinically and biologically important drug-resistant or drug-sensitive genes, where genes were significant in multiple CRISPR screens and multiple pediatric AML cohorts ( Fig 1b). Six genes were determined to resist multiple drugs, with higher expression associated with worse outcomes across multiple cohorts. Particularly, ABCC1 encodes the ATP Binding Cassette C1-mediated drug efflux transporter; therefore, knocking out the efflux transporter causing drug resistance, would increase the lethality of AML cells, resulting in poor outcomes in multiple cohorts in patients treated with ADE alone and ADE with GO or bortezomib. Five genesshowed significant as sensitive genes for multiple drugs in our CRISPR screens with high expression association with favorable outcomes in multiple cohorts. Interestingly, two genes were identified as bortezomib-sensitive genes with high expression in patients treated with ADE+bortezemib in AAML1031 cohort had better outcomes. Other genes were revealed to be sensitive to one drug but resistant to another, indicating that antagonistic pleiotropy effects between genes and drugs may affect clinical outcomes. Current research focuses on developing a pipeline to predict different drug combination outcomes for prioritizing drug treatments and identifying novel drug targets to overcome drug resistance to ADE+GO or bortezomib.

Supplementary Methods, figures and tables

Cells pre-treated for 3h with 20 µM nelfinavir.

Tissue iron overload is a frequent pathologic finding in multiple disease states including non-alcoholic fatty liver disease (NAFLD), neurodegenerative disorders, cardiomyopathy, diabetes, and some forms of cancer. The role of iron, as a cause or consequence of disease progression and observed phenotypic manifestations, remains controversial. In addition, the impact of genetic variation on iron overload related phenotypes is unclear, and the identification of genetic modifiers is incomplete. Here, we used the Hybrid Mouse Diversity Panel (HMDP), consisting of over 100 genetically distinct mouse strains optimized for genome-wide association studies and systems genetics, to characterize the genetic architecture of dietary iron overload and pathology. Dietary iron overload was induced by feeding male mice (114 strains, 6-7 mice per strain on average) a high iron diet for six weeks, and then tissues were collected at 10-11 weeks of age. Liver metal levels and gene expression were measured by ICP-MS/ICP-AES and RNASeq, and lipids were measured by colorimetric assays. FaST-LMM was used for genetic mapping, and Metascape, WGCNA, and Mergeomics were used for pathway, module, and key driver bioinformatics analyses. Mice on the high iron diet accumulated iron in the liver, with a 6.5 fold difference across strain means. The iron loaded diet also led to a spectrum of copper deficiency and anemia, with liver copper levels highly positively correlated with red blood cell count, hemoglobin, and hematocrit. Hepatic steatosis of various severity was observed histologically, with 52.5 fold variation in triglyceride levels across the strains. Liver triglyceride and iron mapped most significantly to an overlapping locus on chromosome 7 that has not been previously associated with either trait. Based on network modeling, significant key drivers for both iron and triglyceride accumulation are involved in cholesterol biosynthesis and oxidative stress management. To make the full data set accessible and useable by others, we have made our data and analyses available on a resource website. Author summary The response to a high iron diet is determined in part by genetic factors. We now report the responses to such a diet in a diverse set of inbred strains of mice, known as the Hybrid Mouse Diversity Panel, that enables high resolution genetic mapping and systems genetics analyses. The levels of iron in the liver varied about >5 fold across the strains, with genetic variation explaining up to 74% of the variation in liver iron. Pathologies included copper deficiency, anemia, and fatty liver, with liver triglycerides varying over 50 fold among the strains. Genetic mapping and network modeling identified significant genetic loci and pathways underlying the response to diet.