BackgroundAcute myeloid leukemia(AML)is a malignant clonal disease originating from the hematopoietic system.The "7+3" induction chemotherapy with cytarabine and anthracycline drugs,followed by high-dose cytarabine consolidation therapy after remission,remains the first-line treatment for newly diagnosed AML patients except acute promyelocytic leukemia(APL).However,due to the high heterogeneity of AML,there are significant differences in the therapeutic effects of the "7+3" regimen among different patients.In recent years,molecular targeted drugs that have emerged one after another are only effective for specific gene types of patients.Therefore,there is still an unmet need for new treatments for AML.Abnormal epigenetic regulation plays an important role in the occurrence and development of AML.Through research on epigenetic modifications in AML,it has been found that epigenetic regulatory molecules closely related to its occurrence and development may become potential targets for intervening in AML.Histone modification is an important epigenetic modification.Since its discovery in the 1960s,a large amount of research has reported its role in cancer.Our team previously screened a compound library and found that UNC0379,a histone methyltransferase SETD8 inhibitor could achieve more than 90%inhibition rate in three non-APL AML cell lines;therefore we locked SETD8 as our study object.Data analysis showed that SETD8 was highly expressed in samples from AML patients and its overexpression was associated with poor prognosis indicating its correlation with the occurrence and development of AML.Previous studies have shown that abnormal expression of SETD8 can promote or inhibit different tumors’ progression depending on their type;however,since SETD8’s role and regulatory mechanism during the occurrence and development of AML remain unclear clarifying this will provide evidence-based intervention strategies based on targeting SETD8 for AML,especially non-APL AML.In recent years,precise treatment has become a trend in the treatment of AML.The basis of precision treatment is accurate prediction of patients’ prognosis.Currently,the European Leukemia Net(ELN)risk stratification based on gene and chromosome mutations is mainly used to predict the prognosis of AML;however,this risk stratification does not consider the potential impact of heterogeneity in gene expression levels at transcriptome level on patient prognosis.Therefore,it is necessary to construct a comprehensive scoring system that includes gene expression information to personalize prognostic predictions for patients.Lately,with the rapid evolution of both theory and technology in bioinformatics,the amount of omics big data such as transcriptomics and genomics generated from gene chips and second-generation sequencing has increased exponentially.Therefore,utilizing bioinformatics methods to analyze this type of big data has significant advantages in establishing accurate prognostic prediction models and formulating personalized treatment plans with broad application prospects.Ferroptosis is a common programmed cell death pathway characterized by unique morphological features such as mitochondrial shrinkage accompanied by decreased cristae density leading to outer membrane rupture.The main molecular mechanisms underlying ferroptosis include intracellular iron overload,increased reactive oxygen species(ROS),lipid peroxidation,depletion of intracellular reducing compound glutathione(GSH),etc.,Since Scott J Dixon proposed ferroptosis in 2012 irondependent cell death-related genes have gained widespread attention due to their important roles played in metabolic diseases cardiovascular diseases neurological disorders as well as tumors.Clarifying the value of ferroptosis-related genes in predicting prognosis for AML will be crucially important for personalizing prognostic predictions and treatments for AML patients.Purpose1.Elucidating the role and mechanism of histone methyltransferase SETD8 in the development of AML.2.Based on the AML public dataset,a multivariate proportional hazard regression(also known as COX)model was constructed using bioinformatics methods to predict patient prognosis with iron death-related genes.Methods and Results1.Downregulation of SETD8 can lower the level of H4K20mel methylation in the CXCR4 promoter region,thereby reducing CXCR4 expression and inhibiting AML cell proliferation while inducing apoptosis.(1)Through screening of a small molecule compound library in the early stage,we identified UNC0379,a selective inhibitor of histone methyltransferase SETD8,which can inhibit the proliferation of three types of AML cell lines(U937,THP1,KASUMI1)with an inhibition rate greater than 90%.Bioinformatics analysis and specimen detection found that SETD8 is highly expressed in newly diagnosed AML patients and is associated with poor prognosis.These results suggest that SETD8 may be a candidate target for AML intervention.(2)We constructed stable leukemia cells U937 and Kasumi-1 with knocked down SETD8.CCK-8 and EdU experiments showed that knocking down SETD8 significantly inhibited the proliferation of AML cells.Flow cytometry detected that knocking down SETD8 promoted apoptosis in AML cells.Western blot experiments showed that after knocking down SETD8,anti-apoptotic protein Bcl-2 levels decreased while apoptotic activation markers such as cle-caspase3 and cle-parpl increased in AML cells.This indicates that inhibiting SETD8 plays a role by inducing cell apoptosis.Through screening death signal pathway-related inhibitors,it was found that only ZVAD-FMK related to apoptotic signaling pathway could restore the growth inhibition caused by SETD8.(3)Using p53 inhibitor Pifithrin-alpha to treat SETD8 knockdown AML cell lines revealed that inhibiting p53 cannot reverse the growth inhibition effect on AML cells or promote apoptosis caused by SETD8 knockdown;indicating SETD8 exerts its effects through non-p5 3-dependent apoptotic pathways.(4)CCK-8 experiments were conducted at concentrations ranging from 1 to 10 μM UNC0379 for half-maximal inhibitory concentration(IC50)screening in AML cells;it was determined that UNC0379 had an IC50 of 7 μM(U937)and 4 μM(Kasumi-1),respectively.Using this concentration,it was verified that UNC0379 can significantly inhibit the proliferation of AML cells and promote apoptosis.(5)We used C1498 cell line to construct a mouse model of AML.Injecting SETD8 knockdown cells or treating mice with SETD8 inhibitor UNC0379 slowed down leukemia progression in vivo,reduced infiltration of leukemia cells in visceral organs,and prolonged survival time.(6)RNA-seq analysis was performed on U937 cells with knocked down SETD8 and untreated U937 cells;enrichment analysis found that differentially expressed genes were involved in multiple signaling pathways related to apoptosis and cancer.Combining TCGA and BeatAML public databases with genes positively correlated with SETD8 expression revealed CXCR4 as a significant gene showing positive correlation with SETD8 expression among intersected genes from self-test RNA-seq differential genes.Further validation using Western Blotting and PCR showed that knocking down SETD8 could lead to decreased mRNA and protein levels of CXCR4;indicating CXCR4 may be a downstream target molecule for SETD8.(7)Through WASHU website search,we found an enriched area of histone H4K20me1 within the upstream 2kb region of the CXCR4 promoter.Western blotting detected that knocking down SETD8 or adding its inhibitor UNC0379 could reduce H4K20me1 protein levels in AML cells;Chip-qPCR results showed that knocking down SETD8 or using its inhibitor UNC0379 significantly reduced the enrichment level near H4K20me1 at the upstream region adjacent to the CXCR4 promoter compared to control group.This indicates that SETD8 regulates transcriptional expression by modulating H4K20me1 levels at the CXCR4 promoter region.(8)Through CCK-8 and EdU recovery experiments,it was found that overexpression of CXCR4 or using its agonist SDF-1α can reverse the inhibitory effect of SETD8 knockdown on cell proliferation in U937 and Kasumi-1 cells.Flow cytometry detected that using SDF-1a in SETD8 knockdown cell lines also led to a decrease in apoptosis levels in AML cells.This further confirms that SETD8 exerts its biological effects by regulating CXCR4.2.Integrating public dataset and ferroptosis gene set,we constructed molecular subtypes of AML related to ferroptosis genes.Based on this,we built an AML prognostic model using 13 differentially expressed genes associated with these subtypes.We also created a prognosis-related column chart by incorporating patient age and mutation-related information.(1)Univariate COX regression was performed on 259 ferroptosis-related genes in the TCGA dataset.Forty-seven genes with prognostic significance were selected for further analysis,and patients in the TCGA dataset were divided into two subtypes,C1 and C2,based on gene expression data using unsupervised clustering.The consensus CDF curve,Nbclust algorithm,and PCA analysis confirmed significant differences between the two subtypes.Using intra-group proportion algorithm,the distribution of similar C1 and C2 subtypes in GEO merged dataset was validated.(2)The clinical baseline data differences between the two subtypes of C1 and C2 were compared in the TCGA dataset.More samples in group C1 belonged to a subgroup with good cytogenetic risk while group C2 had a higher proportion of FAB subtype M5 patients but fewer M3 patients with significantly elevated white blood cell levels.Further analysis of both subtypes in TCGA and GEO merged datasets showed that subtype C2 exhibited poor prognosis according to survival analysis.Tumor infiltration analysis showed that subtype Cl had a higher percentage of CD8+T cells,tumorinfiltrating lymphocytes(TILs),stronger type Ⅰ interferon response as well as lower percentages of regulatory T cells than subtype C2 did.Mutation analysis revealed that RUNX1 mutation frequency was higher in C1 while FLT3-ITDand NPM1 mutations occurred more frequentlyinC2.GSVAanalysis found that promoting cancer signaling pathways such as ERBB,TGF-β,and KRAS were upregulated while cancer suppressive pathways including P53and G2M checkpoints were downregulated in subtype C2;selenoamino acid metabolism pathway was upregulated while peroxisome pathway was downregulated in subtype C2.(3)Using C1 and C2 subtypes as groups,differential genes were identified between the two groups.Lasso-COX joint analysis was performed on these differential genes to obtain a multivariate COX prognostic risk model consisting of 13 genes.Scores were calculated based on the expression values of these 13 genes and their corresponding risk coefficients,and its predictive ability was validated by time-dependent ROC curves in the training set and multiple validation sets.Compared with three reported AML prognostic models,this risk score had better predictive ability.Pan-cancer analysis showed that this risk score also had good prognostic indications for diffuse large B-cell lymphoma,cervical squamous cell carcinoma and endometrial adenocarcinoma,melanoma as well as thyroid cancer.(4)Multiple clinical baseline indicators and risk scores were included,and through multivariate COX regression analysis,it was found that the risk score,age,FLT3-ITD,RUNX1 and TP53 are independent prognostic factors.By integrating age,relevant mutations and risk scores,we constructed a column chart which showed good predictive ability for the survival of AML patients diagnosed at 1 year,2 years and 3 years.The column chart was trained using TCGA as the training set and BeatAML as the external validation set.Both sets demonstrated good predictive ability with high accuracy.(5)After dividing patients into high-risk and low-risk groups based on their respective scores,differences between two groups’ baseline characteristics were compared;significant differences existed in terms of onset age,bone marrow primitive cell percentage,peripheral blood white blood cell count,FAB subtype,cytogenetic,molecular classification risks etc.Higher expression levels of anti-ferroptosis gene GPX4were observed among high-risk group patients who carried FLT3-ITDand DNMT3 Amutations with higher scores.Comparing immune features between high-and low-risk groups revealed higher CD8+T cells,costimulatory T cells,TILs,and type ⅠIFN response among low-risk group while regulatory T cell levels increased among high-risk group.(6)Using CTRP drug sensitivity database,it was found that traditional cytotoxic drugs such as cytarabine,doxorubicin,and etoposide as well as BCL-2 inhibitor venetoclax were more sensitive in low-risk group while demethylating agents(HMAs)such as decitabine and azacitidine as well as histone deacetylase inhibitors(HDACi)such as vorinostat were more sensitive in high-risk group.In addition,it was found that the high expression of GPX4 among high-risk groups had higher sensitivity to RSL3and ML210 which inhibited GPX4-induced ferroptosis.These results were further validated in the human TCGA dataset.Conclusions1.This study found that the expression of histone methyltransferase SETD8 is upregulated in AML and is associated with poor prognosis.SETD8 can enhance CXCR4 transcription by increasing the methylation level of H4K20me1 in the CXCR4 promoter region,promoting AML cell proliferation while inhibiting apoptosis,thereby promoting the occurrence and development of AML.This study provides a basis for developing intervention strategies for AML targeting SETD8.2.In this study,two subtypes related to ferroptosis were constructed using public data on AML and ferroptosis gene sets.Based on this,a 13-gene risk model related to ferroptosis was established,risk scores were calculated,and risk groups were determined.The risk score showed good prognostic prediction ability in multiple external validation sets as well as comparisons with several reported AML prognostic models.The risk group can also predict drug sensitivity and guide disease treatment. |