Abstract
Background:
Most transcriptomic studies in acute myeloid leukemia (AML) have focused on transcriptional regulation, whereas the clinical and biological relevance of translation initiation factors remains insufficiently defined. Eukaryotic translation initiation factor 2 subunit alpha (EIF2S1) is a key regulator of translation initiation, but its prognostic significance and association with AML remain unclear.
Methods:
Bulk RNA-seq data from the TCGA-LAML and GTEx cohorts were analyzed to evaluate EIF2S1 expression, prognostic value, associated biological programs, ssGSEA-derived immune-cell-associated signature scores, and quanTIseq-estimated immune-cell proportions. Single-cell RNA sequencing data were further examined to characterize the cellular distribution of EIF2S1 and subset-specific immune and metabolic transcriptional features. Finally, siRNA-mediated EIF2S1 knockdown assays were performed in THP-1 monocytic AML cells to assess its functional relevance.
Results:
EIF2S1 was significantly upregulated in AML and independently associated with shorter overall survival in the TCGA-LAML cohort. High EIF2S1 expression was linked to translation initiation-related programs and immune- and myeloid-lineage-related transcriptional features, including stronger myeloid-lineage signals, weaker cytotoxic T-cell-related signatures, and increased expression of immune checkpoint and immunoregulatory genes, including LGALS9 and TGFB1. Single-cell analysis localized EIF2S1 enrichment primarily to monocyte-like subsets and associated it with mitochondrial metabolic and innate immune transcriptional programs. Functionally, EIF2S1 knockdown induced G0/G1 cell-cycle arrest, suppressed THP-1 cell proliferation, migration, and invasion, and promoted apoptosis.
Conclusions:
EIF2S1 is associated with adverse prognosis, translation initiation-related programs, and monocyte-like immune-metabolic transcriptional states in AML. Functional findings further suggest that EIF2S1 contributes to leukemic cell fitness. These results support EIF2S1 as a potential therapeutic target requiring further experimental and clinical validation.
1 Introduction
Acute myeloid leukemia (AML) is a highly heterogeneous hematologic malignancy that continues to be associated with poor clinical outcomes (). Extensive genomic and transcriptomic studies have identified recurrent mutations, aberrant signaling pathways, and distinct molecular subtypes in AML; however, these alterations do not fully explain the biological complexity and clinical variability of the disease (, ). These limitations highlight the need to investigate additional regulatory layers beyond transcription.
Translation initiation represents the rate-limiting step of protein synthesis, and its dysregulation is a fundamental driver in the pathogenesis of various malignancies, including hematopoietic tumors (, ). Eukaryotic translation initiation factor 2 subunit alpha (EIF2S1) is a central component of the translation initiation machinery that participates in the delivery of initiator methionyl-tRNA to the ribosome (). While burgeoning evidence in solid tumors suggests that EIF2S1 activation supports cell survival, metabolic reprogramming, and chemoresistance under microenvironmental stress (, ). Importantly, total EIF2S1 expression and Ser51 phosphorylation of eIF2α represent related but distinct regulatory layers: total expression reflects the abundance of a translation-initiation component, whereas Ser51 phosphorylation is a key regulator of integrated stress response activity and translational output. Thus, EIF2S1 expression should not be interpreted as equivalent to its phosphorylation status or pathway activity (, ). However, the clinical significance and biological relevance of EIF2S1 expression in AML remain insufficiently characterized.
The bone marrow microenvironment (BMM) of AML is a unique immunosuppressive microenvironment characterized by polarized M2-like macrophages and a depleted T-cell population (–). This malignant cellular subset must finely regulate its translational output to balance high metabolic demands with the continuous secretion of potent immunomodulatory factors (, ). Immunomodulatory factors secreted by tumor cells are highly dependent on fine-tuning at the translational level (). As a core component of the translation initiation machinery, EIF2S1 may be involved in linking translational control with cellular stress responses, metabolic adaptation, and immune-related transcriptional programs (, ). However, the cell-type-specific distribution of EIF2S1 and its associations with the AML immune-metabolic landscape remain unclear.
Accordingly, this study investigated the dual clinical and biological relevance of EIF2S1 in AML. Bulk RNA-seq data from TCGA-LAML and GTEx were used to evaluate EIF2S1 expression, prognostic significance, and associated molecular programs. ssGSEA was used to quantify relative immune-cell-associated signature scores, whereas quanTIseq was used to computationally estimate immune-cell proportions. Single-cell RNA-sequencing analysis was performed to characterize the cellular distribution of EIF2S1 and its associations with immune and metabolic transcriptional states. Finally, we performed EIF2S1 loss-of-function assays in the monocytic AML cell line THP-1, selected in view of the monocyte-like states identified by single-cell analysis, to evaluate its contribution to leukemic cell fitness, including proliferative capacity, cell-cycle progression, migration, and invasion. Collectively, this study investigated whether EIF2S1 serves as a prognostic marker associated with monocyte-related immune-metabolic states and contributes to leukemic cell fitness.
2 Materials and methods
2.1 Data acquisition
Bulk RNA sequencing (RNA-seq) expression profiles and matched clinical follow-up data for The Cancer Genome Atlas acute myeloid leukemia cohort (TCGA-LAML) were downloaded from the UCSC Xena platform. Normal whole-blood samples were obtained from the Genotype-Tissue Expression project (GTEx); samples without clinical information were excluded. TCGA-LAML and GTEx expression data were obtained from the UCSC Xena Toil RNA-seq recompute compendium, in which both datasets were processed using a unified computational pipeline and provided as RSEM-normalized transcripts per million (TPM) values. Expression values were transformed as log2(TPM + 1) before analysis. Risk stratification was assigned according to European LeukemiaNet (ELN) criteria (). Single-cell RNA sequencing (scRNA-seq) data were obtained from Gene Expression Omnibus (GEO; GSE154109), including 8 AML patients and 4 healthy controls.
2.2 Bulk expression, survival, and co-expression analyses
EIF2S1 expression between AML and controls was compared using the Mann-Whitney U test. Patients were dichotomized into EIF2S1-high and EIF2S1-low groups using the median expression as the cutoff. Overall survival (OS) was analyzed by Kaplan-Meier method and compared by log-rank test. Cox proportional hazards models were fitted using the R survival package for univariate and multivariate analyses to evaluate the independent association of EIF2S1 with OS. Genome-wide co-expression with EIF2S1 was assessed using Spearman correlation in the TCGA-LAML cohort.
2.3 Protein-protein interaction network
A protein-protein interaction (PPI) network for EIF2S1 was generated using STRING database () with a high-confidence threshold (overall score ≥0.900). Network visualization was performed using circlize to generate Circos plots.
2.4 Differential expression and enrichment analyses
Differential expression analysis between the EIF2S1-high and EIF2S1-low groups was performed using the limma package. EIF2S1-high and EIF2S1-low cases were defined using the cohort median. A linear model was fitted to the log2(TPM + 1)-transformed expression matrix, and empirical Bayes moderation was applied to obtain moderated t-statistics. Genes with an absolute log2 fold change greater than 1 and a Benjamini-Hochberg-adjusted p value below 0.05 were considered differentially expressed. Gene Ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) enrichment analyses were performed using clusterProfiler ().
2.5 Assessment of immune-cell-related transcriptional features
Immune-cell-related transcriptional features were assessed using two complementary computational approaches. Single-sample gene set enrichment analysis (ssGSEA), implemented in the GSVA package, was used to calculate relative enrichment scores for predefined immune-cell-associated genesignatures (). quanTIseq deconvolution was used to computationally estimate the relative proportions of predefined immune-cell populations (, ). ssGSEA scores and quanTIseq estimates were interpreted as computationally inferred features rather than direct measurements of immune-cell infiltration. To assess potential confounding by monocytic differentiation, sensitivity analyses were performed after excluding FAB M4/M5 cases; detailed procedures are provided in Supplementary Methods 2.
2.6 scRNA-seq processing and downstream analyses
scRNA-seq data were analyzed using Seurat (). Low-quality cells and putative doublets were removed by stringent filtering: cells with fewer than 200 or more than 5,000 detected genes (nFeature_RNA), or a total UMI count (nCount_RNA) exceeding 40,000, were excluded. We also removed cells with mitochondrial gene content (percent.mt) > 5%, hemoglobin gene expression (percent.hb) > 0.1%, or ribosome gene expression (percent.ribo) > 30%. After quality control, 3,346 cells were retained for downstream analysis. Data were normalized using SCTransform, followed by principal component analysis (PCA). Batch effects were corrected using Harmony () based on the first 22 principal components. The first 22 Harmony dimensions were used for nearest-neighbor graph construction and UMAP visualization, with n.neighbors = 30 and min.dist = 0.3. Louvain clustering was performed using the Seurat FindClusters function at a resolution of 2, yielding 23 clusters. Clusters 0–22 contained 344, 288, 286, 246, 233, 215, 201, 200, 172, 157, 156, 145, 141, 135, 107, 101, 62, 34, 27, 26, 25, 23, and 22 cells, respectively. Cell types were annotated using canonical lineage markers curated in CellMarker 2.0 (). The candidate marker panels included monocytes (CD14, LYZ, S100A8, S100A9, VCAN, and FCGR3A), CD8+ effector memory RA cells (CD3D, CD8A, GZMA, GZMB, PRF1, NKG7, and KLRG1), plasmacytoid dendritic cells (LILRA4, CLEC4C, IL3RA, IRF7, and TCF4), naive B cells (CD79A, MS4A1, IGHD, TCL1A, FCER2, and IL4R), myeloid dendritic cells (CD1C, FCER1A, CLEC10A, HLA-DQA1, and HLA-DQB1), class-switched mature B cells (CD79A, MS4A1, CD27, IGHG1, and IGHA1), and common myeloid progenitors (CD34, KIT, FLT3, IL3RA, CSF1R, and GATA2). The final annotated populations comprised 1,622 monocytes, 825 CD8+ effector memory RA cells, 698 naive B cells, 141 common myeloid progenitors, and 60 class-switched mature B cells. We divided the monocyte subsets into an EIF2S1 positive (detected expression > 0) group and an EIF2S1 negative (detected expression = 0) group. DEGs were identified using the Wilcoxon rank-sum test, followed by GO and KEGG enrichment analyses.
2.7 Cell culture and siRNA-mediated EIF2S1 knockdown
The human acute monocytic leukemia cell line THP-1 (Cat. No. CL-0233) was purchased from Wuhan Procell Biotechnology Co., Ltd. THP-1 cells were selected as a monocytic AML model because the single-cell analysis associated EIF2S1 expression with monocyte-like leukemic cell states. The experiments were intended to provide initial functional validation in this specific cellular context. Cells were cultured in RPMI-1640 supplemented with 10% fetal bovine serum (FBS; Gibco) and 1% penicillin–streptomycin at 37 °C with 5% CO2. Three EIF2S1-targeting small interfering RNAs (siRNAs: si1#, si2#, si3#) and a negative control (si-NC) were used. The siRNA sequences targeting EIF2S1 were as follows: negative control sense strand: 5’-UUCUCCGAACGUGUCACGUTT-3’, antisense strand: 5’-ACGUGACACGUU CGGAGAATT-3’; siRNA-1 sense strand: 5’-GGCUUGUUAUGGUU AUGAATT-3’, antisense strand: 5’-UUCAUAACCAUAACAAGCCTT-3’; siRNA-2 sense strand: 5’-GUCAAGCUAUGGCUGUUAUTT-3’, antisense strand: 5’-AUAACAGCCAUA GCUUGACTT-3’. siRNA-3 sense strand: 5’-GCUUGCU GGAAUACAACAATT-3’, antisense strand:5’-UUGUUGUAUUCCAGCAAGCTT-3’. Cells were transfected with siRNA-Mate SUS according to the manufacturer’s protocol and harvested 48–72 h post-transfection. Knockdown efficiency was assessed at mRNA and protein levels by real-time quantitative PCR (RT-qPCR) and Western blotting to select the optimal siRNA for functional assays.
2.8 RT-qPCR and Western blotting
Total RNA was extracted using TRIzol (Invitrogen) and reverse-transcribed to cDNA. RT-qPCR was performed using Hieff® qPCR SYBR® Green Master Mix (Low ROX); β-actin was used as the internal control. The primer sequences were as follows: β-actin: Forward: GACAGGATGCAGAAGGAGAT;Reverse: GAGGCCAGGATGGAGC;EIF2S1:Forward: TGGTGAATGTCAGATCCATTGC;Reverse: TAGAACGGATACGCCTTCTGG. For Western blotting, total protein was extracted using RIPA buffer with protease inhibitors, quantified by BCA assay, separated by SDS-PAGE, and transferred to PVDF membranes. Membranes were blocked with 5% non-fat milk, incubated with primary antibodies (anti-EIF2S1 1:2,000; anti-β-actin 1:2,000) overnight at 4 °C, followed by HRP-conjugated secondary antibodies for 1 h at room temperature. Signals were detected using ECL.
2.9 Functional assays
Cell proliferation was measured by Cell Counting Kit-8 (CCK-8; Cat. No. K1018) at 0, 24, 48, 72 and 96 h, with absorbance read at 450 nm after 4 h incubation. Cell-cycle distribution was determined by propidium iodide (PI) staining with RNase (Cat. No. K2263) and analyzed by flow cytometry (FongCyte). Apoptosis was assessed using an Annexin V-FITC/PI apoptosis detection kit (Cat. No. K2003) according to the manufacturer’s instructions, followed by flow-cytometric analysis. Migration and invasion were assessed using Transwell assays; invasion inserts were pre-coated with Matrigel (Cat. No.C231001). After 24 h, migrated/invaded cells were fixed, stained with 0.1% crystal violet, and quantified in randomly selected fields.
2.10 Statistical analysis
All analyses were performed in R (v4.5.2) and GraphPad Prism (v9.5). Quantitative data are presented as the mean ± standard deviation (SD) of three independent biological experiments. Because only three biological replicates were available per condition, formal normality testing was not performed. qRT-PCR, immunoblot densitometry, migration, invasion, and apoptosis data were analyzed using one-way ANOVA followed by Dunnett’s multiple-comparison test. Time-course CCK-8 data and cell-cycle distributions were analyzed using two-way ANOVA followed by Dunnett’s multiple-comparison test. All tests were two-sided, and P < 0.05 was considered statistically significant. p < 0.05 was considered statistically significant (*p < 0.05; **p < 0.01; ***p < 0.001; ****p < 0.0001).
3 Results
3.1 EIF2S1 is significantly overexpressed in AML and is positively correlated with poor patient prognosis
To investigate the clinical significance of EIF2S1 in AML, we first compared its expression levels between AML patients and normal controls using public databases (TCGA and GTEx). Differential expression analysis revealed that EIF2S1 was significantly overexpressed in AML patients compared to normal (Figure 1A; p < 0.0001). To reduce the potential influence of cross-project differences between TCGA-LAML and GTEx, EIF2S1 expression was independently evaluated in the GSE30029 cohort using cell-type-matched CD34-positive samples. EIF2S1 expression was significantly higher in AML CD34-positive cells than in normal bone marrow CD34-positive cells (n = 46 versus n = 31; p = 0.005913; Supplementary Figure 1), supporting the reproducibility of EIF2S1 upregulation in AML.
Figure 1
Subsequently, Kaplan-Meier analysis subsequently showed that patients in the EIF2S1-high group had significantly shorter overall survival than those in the EIF2S1-low group (Figure 1B; HR = 1.51, 95% CI: 1.02–2.22, p = 0.038). Consistently, the distribution of survival outcomes indicated a survival advantage in patients with lower EIF2S1 expression (Figures 1C, D). Univariable Cox regression analysis showed that age >60 years, poor cytogenetic risk, and high EIF2S1 expression were associated with adverse overall survival. After adjustment for sex, age, white blood cell count, cytogenetic risk, FLT3 mutation status, and NPM1 mutation status, high EIF2S1 expression remained significantly associated with an increased risk of death (HR = 1.600, 95% CI: 1.085–2.360, p = 0.018; Figure 1E). The prognostic association of EIF2S1 was further evaluated in the independent Beat AML 2.0 cohort. Higher continuous EIF2S1 expression was significantly associated with an increased risk of death after multivariable adjustment (adjusted HR per 1-SD increase = 1.222, 95% CI: 1.032–1.447, p = 0.020; Supplementary Figure 2). In the complementary dichotomized multivariable analysis, the EIF2S1-high group also showed a higher adjusted mortality risk (adjusted HR = 1.37, 95% CI: 1.01–1.86, p = 0.040; Supplementary Figure 3).
As a complementary analysis in the TCGA-LAML cohort, EIF2S1 was also modeled as a continuous standardized variable. Each one-standard-deviation increase in EIF2S1 expression was significantly associated with a higher risk of death after adjustment for sex, age, white blood cell count, cytogenetic risk, FLT3 mutation status, and NPM1 mutation status (adjusted HR = 1.476, 95% CI: 1.110–1.962, p = 0.007; Supplementary Figure 4). Collectively, these findings support EIF2S1 as a candidate prognostic marker associated with adverse survival outcomes in AML.
3.2 EIF2S1 is co-expressed with translation initiation components in AML
To elucidate the molecular interactome landscape of EIF2S1 in AML, a high-confidence PPI network (combined score ≥ 0.900) was first constructed using the STRING database. EIF2S1 was connected to several components of the translation initiation machinery, including EIF2S3, EIF2B2, EIF2B3, EIF3A, and EIF3B, as well as the ribosomal proteins RPS3 and RPS3A (Figure 2A). Correlation analysis in the TCGA-LAML cohort further revealed coordinated expression among these translation-related genes (Figure 2B).
Figure 2
To define the broader transcriptional context associated with EIF2S1, AML samples were ordered according to EIF2S1 expression, and the top 20 EIF2S1-correlated genes were visualized in a heatmap (Figure 2C; p < 0.001). All selected genes were positively correlated with EIF2S1, with correlation coefficients ranging from 0.708 to 0.789. Their expression progressively increased across samples with higher EIF2S1 levels, indicating that EIF2S1 upregulation occurs within a coordinated transcriptional program rather than as an isolated expression change. Collectively, these findings place EIF2S1 within a translation initiation-associated molecular network in AML.
3.3 EIF2S1-high AML exhibits distinct immune and metabolic transcriptional programs
Transcriptomic profiling comparing EIF2S1-high and EIF2S1-low AML cohorts identified a total of 373 DEGs, including 276 upregulated and 97 downregulated genes, using limma with empirical Bayes moderation and the predefined thresholds of |log2 fold change| > 1 and adjusted p < 0.05 (Figure 3A). Ranked-list gene set enrichment analysis (GSEA) showed positive enrichment of immune-related programs in the EIF2S1-high group, including antigen processing and presentation, MHC class II protein complex assembly, and cellular defense responses. Additional inflammation- and immune-response-related pathways were also enriched (Figures 3B, C). Conversely, negatively enriched pathways predominantly involved metabolic and regulatory processes, including ascorbate and aldarate metabolism and spliceosome function (Figure 3D). GO analysis of the DEG set identified enrichment in inflammatory-response regulation, leukocyte migration, responses to bacterial-origin molecules and lipopolysaccharide, membrane- and secretory-granule-associated cellular components, and immune- and pattern-recognition-receptor activities (Figure 3E). KEGG analysis further identified enrichment in phagocytosis, neutrophil extracellular trap formation, neuroactive ligand-receptor interaction, and selected metabolic and pathogen-response-labeled pathways (Figure 3F). The pathogen-response-labeled pathways were interpreted as reflecting shared innate immune, inflammatory, and phagocytic gene modules rather than evidence of specific infectious processes. Collectively, these findings associate EIF2S1-high AML with antigen-presentation-, inflammatory-, innate immune-, leukocyte-migration-, and phagocytosis-related transcriptional programs, together with alterations in selected metabolic and RNA-processing pathways.
Figure 3
3.4 EIF2S1 expression is associated with myeloid- and T-cell-related transcriptional features in AML
To characterize immune-cell-related transcriptional features associated with EIF2S1 expression in AML, we applied two complementary computational approaches, ssGSEA and quanTIseq. ssGSEA analysis showed that continuous EIF2S1 expression was positively correlated with macrophage- and neutrophil-related signature scores, whereas inverse correlations were observed with total T-cell-, CD8-positive T-cell-, and cytotoxic-cell-related signature scores (Figures 4A, B). Complementary quanTIseq analysis estimated higher relative proportions of M2-like macrophages and myeloid dendritic cells in the EIF2S1-high group (Figures 4C, D). Because these analyses were based on bulk RNA-sequencing data, the findings were interpreted as immune- and myeloid-lineage-related transcriptional features rather than direct evidence of altered immune-cell infiltration. To examine the potential influence of monocytic AML composition, we performed sensitivity analyses according to FAB subtype. EIF2S1 expression did not differ significantly between FAB M4/M5 and non-M4/M5 cases. After exclusion of FAB M4/M5 cases, the positive association between EIF2S1 and the macrophage-related ssGSEA score remained evident. Several additional myeloid- and T-cell-related associations remained directionally consistent. The quanTIseq analysis also continued to show positive correlation patterns between EIF2S1 and selected myeloid-cell estimates, including M2-like macrophages and myeloid dendritic cells (Supplementary Figure 5).
Figure 4
3.5 EIF2S1 correlates with immune checkpoint-related genes and heterogeneous checkpoint expression patterns
In the TCGA-LAML cohort, EIF2S1 expression was significantly positively correlated with multiple immune checkpoint-related genes, including CD274, CD40, CD48, IL10, LGALS9, PDCD1LG2 and TGFB1 (Figure 5A; all p < 0.05), suggesting a potential role for EIF2S1 in modulating the immune microenvironment in AML. Notably, differential expression analysis revealed marked heterogeneity in the expression patterns of these immune checkpoint genes in AML. Specifically, CD274, CD48 and IL10 were overall downregulated, whereas LGALS9, PDCD1LG2 and TGFB1 were significantly upregulated (Figures 5B–G; p < 0.0001). These findings indicate that the relationship between EIF2S1 and immune checkpoint regulation may be influenced by distinct molecular backgrounds and regulatory mechanisms within AML.
Figure 5
3.6 Single-cell transcriptomics localizes EIF2S1 enrichment to monocyte-like subsets in AML
Single-cell transcriptome mapping showed that, compared with healthy controls (HC), the immune microenvironment components of AML patients underwent significant remodeling, mainly manifested as an increase in the proportion of monocytes and naive B cells, while a sharp decrease in CD8+ effector memory T cells (Figures 6A, B). Gene expression analysis showed that the expression level of the translation initiation factor EIF2S1 was significantly upregulated in the AML microenvironment (Figure 6C). To explore its potential mechanism, we enriched the co-expression network of EIF2S1 in AML monocytes. GO and KEGG analyses revealed that EIF2S1 expression is closely related to the activation of innate immune/stress responses (such as antiviral defense responses) and mitochondrial metabolism (mitochondrial translation, ATP synthesis coupled with electron transport) (Figures 6D, E). In summary, AML samples exhibited marked cellular composition remodeling. EIF2S1 enrichment in monocyte-like subsets was associated with mitochondrial metabolic programs and innate immune/stress-response signatures.
Figure 6
3.7 EIF2S1 knockdown induces G0/G1-phase accumulation and impairs leukemic cell fitness in THP-1 cells
Given the specific enrichment of EIF2S1 in monocytic subsets identified in our scRNA-seq analysis, we selected THP-1, a representative human acute monocytic leukemia cell line, to further functionally validate. We specifically knocked down EIF2S1 in THP-1 cells using small interfering RNA (siRNA). Western blotting and RT-qPCR analysis confirmed that si1# and si3# were the most efficient and were used in subsequent experiments (Figures 7A, B). Cell kinetic analysis showed that the loss of EIF2S1 significantly inhibited the proliferation rate of THP-1 cells (Figure 7C). Flow cytometry further showed that knockdown of EIF2S1 led to significant G0/G1 phase cell cycle arrest (Figures 7D, E). Furthermore, in vitro experiments confirmed that downregulation of EIF2S1 severely impaired the cell’s motility potential: compared with si-NC, the loss of EIF2S1 greatly inhibited the migration ability of THP-1 cells (Figures 7F, G), and the number of invading cells penetrating the basement membrane was also significantly reduced (Figures 7H, I). EIF2S1 knockdown increased the proportion of apoptotic cells in THP-1 cultures (Supplementary Figure 6; Supplementary Table 1). Collectively, these findings indicate that EIF2S1 knockdown suppresses proliferation, induces G0/G1-phase accumulation, promotes apoptosis, and impairs migration and invasion in THP-1 cells.
Figure 7
4 Discussion
While the application of novel targeted therapies has made AML treatment more personalized, chemotherapy resistance and relapse remain major clinical challenges (, ). Immunotherapy has shown great promise in solid tumors, but its efficacy in AML is heavily limited by the highly immunosuppressive bone marrow microenvironment (, ). Emerging evidence reveals that the establishment of this immunosuppressive niche relies heavily on aberrant protein synthesis and metabolic adaptation (, ). Dysregulated translational control is increasingly recognized as an important mechanism linking cancer cell adaptation, immune regulation, and therapeutic resistance (). At the core of this process is the translation initiation factor EIF2S1, widely recognized as a central regulator controlling the rate-limiting step in cellular protein synthesis (). In this study, by integrating transcriptomic data from large-scale public cohorts with single-cell RNA sequencing (scRNA-seq) and in vitro functional assays, we systematically delineated the expression landscape, clinical prognostic value, and biological impact of EIF2S1 in AML. Our findings suggest that EIF2S1 upregulation is associated with adverse prognosis, translation-initiation programs, monocyte-like immune-metabolic features, and malignant phenotypes in monocytic AML cells.
First, we corroborated the elevated expression of EIF2S1 in AML across independent cohorts and obtained external support for its adjusted association with shorter overall survival. More importantly, multivariate Cox regression analysis revealed that high EIF2S1 expression retains independent prognostic value even after adjusting for traditional factors such as patient age, initial cytogenetic risk, and common FLT3 and NPM1 mutations. Given that traditional genomic stratification sometimes fails to fully capture the dynamic functional status of AML at the transcriptomic and translational levels (–), the evaluation of EIF2S1 could serve as a valuable supplement to existing molecular markers, enabling more precise identification of high-risk patients in clinical practice.
At the molecular level, our interaction network and co-expression analyses placed EIF2S1 within a tightly coordinated translation initiation-associated module in AML. EIF2S1 expression was positively correlated with key components of the eIF2 and eIF3 machinery, including EIF2B3, EIF2S3, and EIF3B, indicating coordinated expression of translation-initiation-related genes in EIF2S1-high AML. Such a transcriptional pattern may reflect the increased biosynthetic demands of rapidly proliferating leukemic cells, which require sustained protein synthesis to support cell-cycle progression and malignant growth. This interpretation is consistent with our functional findings that EIF2S1 knockdown induced G0/G1 cell-cycle arrest, suppressed THP-1 cell proliferation and promoted apoptosis.
Beyond maintaining autonomous cell survival, the crosstalk between leukemia cells and the bone marrow immune microenvironment dictates disease progression (, ). Analysis using limma with empirical Bayes moderation and FDR-controlled thresholds yielded a more conservative DEG set, while the principal enrichment patterns remained broadly consistent. Ranked-list GSEA associated EIF2S1-high AML with antigen-processing, inflammatory, and immune-response programs, whereas negatively enriched pathways included selected metabolic and RNA-processing processes. GO and KEGG analyses further highlighted inflammatory-response regulation, leukocyte migration, innate immune recognition, phagocytosis, and membrane- or secretory-granule-associated functions. Several KEGG categories carried infectious-disease labels; however, these pathways likely reflect shared innate immune, inflammatory, and phagocytic gene modules rather than evidence of specific infectious processes. Crucially, EIF2S1 expression exhibited a positive correlation with multiple immune checkpoints (such as LGALS9, TGFB1, and PDCD1LG2). It is biologically plausible that the synthesis of certain immunomodulatory factors harboring complex mRNA structures heavily depends on the overactivation of translation initiation pathways (). Consistently, ssGSEA and quanTIseq analyses associated higher EIF2S1 expression with stronger myeloid-lineage-related transcriptional signals, including higher macrophage-related signature scores and higher estimated relative proportions of M2-like macrophages and selected myeloid-cell populations, together with weaker cytotoxic CD8+T-cell-related signatures. Selected myeloid-related associations remained after exclusion of FAB M4/M5 cases, suggesting that they were not entirely attributable to monocytic AML composition. Collectively, these findings indicate that EIF2S1-high AML is associated with an immune- and myeloid-lineage-related transcriptional context characterized by increased checkpoint-related gene expression and weaker cytotoxic T-cell-related signatures. To resolve this complex cellular ecosystem at higher resolution, our scRNA-seq analysis characterized the microenvironmental heterogeneity. We found that EIF2S1 was enriched in monocyte-like subsets within AML samples. Pathway analysis indicated that this high EIF2S1 expression is tightly linked to highly active innate immune signaling networks and accelerated mitochondrial metabolic processes (including ATP synthesis and the electron transport chain). These findings suggest that EIF2S1-high monocyte-like subsets exhibit coordinated mitochondrial metabolic and innate immune transcriptional features at the single-cell level.
Consistent with the transcriptomic association between EIF2S1 and monocyte-like AML states, THP-1 knockdown experiments showed that EIF2S1 supports leukemic cell proliferation, cell-cycle progression, migration, and invasion. These findings provide functional evidence for the oncogenic relevance of EIF2S1 in monocytic AML cells.
We acknowledge some limitations of this study. In vitro validation was limited to THP-1 cells, a monocytic AML model, chosen based on the association between EIF2S1 and monocytic leukemia status in our single-cell analysis. All functional conclusions are limited to THP-1 cells. Further investigation is needed using other AML cell lines, primary patient-derived cells, and appropriate in vivo models, combined with other mechanistic experiments, to more comprehensively elucidate the role of EIF2S1. Second, although transcriptomic analyses linked EIF2S1 to translation initiation-related programs, the study did not directly assess ribosome occupancy and translational efficiency. Further studies using ribosome profiling, polysome analysis, proteomics, and direct translation assays are needed to define EIF2S1-dependent translational programs. Third, the pharmacological tractability, therapeutic selectivity, and safety of EIF2S1 targeting were not evaluated. These issues should be examined in genetically diverse AML models, primary samples, normal hematopoietic controls, and appropriate in vivo systems. Finally, although the adjusted prognostic association of EIF2S1 received external support in the Beat AML 2.0 cohort, further validation in additional independent, clinically harmonized, and prospectively collected cohorts is required. Collectively, these studies will be necessary to determine whether EIF2S1 can be developed as a safe and clinically relevant therapeutic target in AML.
In summary, EIF2S1 is upregulated in AML and independently associated with adverse survival outcomes. Integrative bulk and single-cell transcriptomic analyses indicate that EIF2S1-high AML is characterized by translation initiation-related programs, monocyte-like immune-metabolic states, and immunosuppressive transcriptional features. Functional studies further support a role for EIF2S1 in maintaining leukemic cell fitness, as its knockdown in THP-1 cells induces G0/G1 cell-cycle arrest and suppresses proliferation, migration, and invasion. Collectively, these findings identify EIF2S1 as a clinically relevant prognostic biomarker and a potential therapeutic target requiring further experimental validation. Additional studies in genetically diverse AML models, primary patient samples, normal hematopoietic controls, and independent clinical cohorts are needed to clarify its underlying mechanisms, determine its therapeutic selectivity, and establish its translational relevance.
Statements
Data availability statement
The datasets presented in this study can be found in online repositories. The names of the repository/repositories and accession number(s) can be found in the article/Supplementary Material.
Ethics statement
Ethical approval was not required for the studies on animals in accordance with the local legislation and institutional requirements because only commercially available established cell lines were used.
Author contributions
XH: Conceptualization, Investigation, Methodology, Writing – original draft. YH: Conceptualization, Investigation, Methodology, Writing – original draft. WW: Conceptualization, Investigation, Methodology, Writing – original draft. XJ: Formal analysis, Writing – original draft. YL: Formal analysis, Writing – original draft. YX: Writing – review & editing. DL: Writing – review & editing.
Funding
The author(s) declared that financial support was received for this work and/or its publication. This study was supported by the National Natural Science Foundation of China (82072355) and the Special Financial Project of Fujian Province (23SCZZX006).
Acknowledgments
We are grateful to the contributors to the public databases used in this study.
Conflict of interest
The author(s) declared that this work was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.
Generative AI statement
The author(s) declared that generative AI was not used in the creation of this manuscript.
Any alternative text (alt text) provided alongside figures in this article has been generated by Frontiers with the support of artificial intelligence and reasonable efforts have been made to ensure accuracy, including review by the authors wherever possible. If you identify any issues, please contact us.
Publisher’s note
All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.
Supplementary material
The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fonc.2026.1922910/full#supplementary-material
References
1
DiNardoCDErbaHPFreemanSDWeiAH. Acute myeloid leukaemia. Lancet. (2023) 401:2073–86. doi: 10.1016/s0140-6736(23)00108-3
2
CarterJLHegeKYangJKalpageHASuYEdwardsHet al. Targeting multiple signaling pathways: the new approach to acute myeloid leukemia therapy. Signal Transd Targ Ther. (2020) 5:288. doi: 10.1038/s41392-020-00361-x
3
ZhangZHuangJZhangZShenHTangXWuDet al. Application of omics in the diagnosis, prognosis, and treatment of acute myeloid leukemia. biomark Res. (2024) 12:60. doi: 10.1186/s40364-024-00600-1
4
SfakianosAPRavenRMSmithTSunX-MMulroneyTEPizzingaMet al. Inhibiting translation elongation by reducing eIF5A activity induces feedback inhibition of initiation, limiting tumour cell proliferation. Nat Commun. (2025) 16(1):11486. doi: 10.1038/s41467-025-66531-z
5
SunLWangWHanCHuangWSunYFangKet al. The oncomicropeptide APPLE promotes hematopoietic Malignancy by enhancing translation initiation. Mol Cell. (2021) 81:4493–508:e4499. doi: 10.1016/j.molcel.2021.08.033
6
LabbéKLeBonLKingBVuNStoopsEHLyNet al. Specific activation of the integrated stress response uncovers regulation of central carbon metabolism and lipid droplet biogenesis. Nat Commun. (2024) 15:8301. doi: 10.1038/s41467-024-52538-5
7
LiYShenYXieMWangBWangTZengJet al. LncRNAs LCETRL3 and LCETRL4 at chromosome 4q12 diminish EGFR-TKIs efficiency in NSCLC through stabilizing TDP43 and EIF2S1. Signal Transd Targ Ther. (2022) 7:30. doi: 10.1038/s41392-021-00847-2
8
ChuangH-YChengC-HGuoFYadavVKFongI-HLinJ-Wet al. Super-enhancer PPP1R15B/EIF2A axis characterizes aggressive drug-tolerant persister cells in glioblastoma. BioMed Pharmacother. (2025) 189:118309. doi: 10.1016/j.biopha.2025.118309
9
NwosuGOPowellJAPitsonSM. Targeting the integrated stress response in hematologic Malignancies. Exp Hematol Oncol. (2022) 11:94. doi: 10.1186/s40164-022-00348-0
10
AmiriMKinirySJPossematoAPMahmoodNBasiriTDufourCRet al. Impact of eIF2α phosphorylation on the translational landscape of mouse embryonic stem cells. Cell Rep. (2024) 43:113615. doi: 10.1016/j.celrep.2023.113615
11
da CruzLCLeeYPaikJYParkJDiederichMCerellaC. Bone marrow tumor microenvironment profiling predicts distinct immunosuppressive phenotypes and immunotherapy potential in AML patients. BioMed Pharmacother. (2025) 189:118287. doi: 10.1016/j.biopha.2025.118287
12
BrauneckFFischerBWittMMuschhammerJOelrichJda Costa AvelarPHet al. TIGIT blockade repolarizes AML-associated TIGIT+ M2 macrophages to an M1 phenotype and increases CD47-mediated phagocytosis. J Immunother Cancer. (2022) 10:e004794. doi: 10.1136/jitc-2022-004794
13
FengJGouJWangYWeiWMaYRenXet al. Bisecting GlcNAc expression by bone marrow stromal cells modulates TGF-β1-driven macrophage polarization in myeloid leukemias. Haematologica. (2025) 110:3009. doi: 10.3324/haematol.2024.287091
14
TangXLiKWangYRocchiSShenSCerezoM. Metabolism and mRNA translation: a nexus of cancer plasticity. Trends Cell Biol. (2025) 35:562–75. doi: 10.1016/j.tcb.2024.10.009
15
HanMHuangQLiXChenXZhuHPanYet al. M7G-related tumor immunity: novel insights of RNA modification and potential therapeutic targets. Int J Biol Sci. (2024) 20:1238. doi: 10.7150/ijbs.90382
16
LiuHZengXRenXZhangYHuangMTanLet al. Targeting tumour-intrinsic N(7)-methylguanosine tRNA modification inhibits MDSC recruitment and improves anti-PD-1 efficacy. Gut. (2023) 72:1555–67. doi: 10.1136/gutjnl-2022-327230
17
ChakrabartyYYangZChenHChanDC. The HRI branch of the integrated stress response selectively triggers mitophagy. Mol Cell. (2024) 84:1090–100:e1096. doi: 10.1016/j.molcel.2024.01.016
18
DangTTKimM-JLeeYYLeHTKimKHNamSet al. Phosphorylation of EIF2S1 (eukaryotic translation initiation factor 2 subunit alpha) is indispensable for nuclear translocation of TFEB and TFE3 during ER stress. Autophagy. (2023) 19:2111–42. doi: 10.1080/15548627.2023.2173900
19
DöhnerHWeiAHAppelbaumFRCraddockCDiNardoCDDombretHet al. Diagnosis and management of AML in adults: 2022 recommendations from an international expert panel on behalf of the ELN. Blood. (2022) 140:1345–77. doi: 10.1182/blood.2022016867
20
SzklarczykDGableALNastouKCLyonDKirschRPyysaloSet al. The STRING database in 2021: customizable protein-protein networks, and functional characterization of user-uploaded gene/measurement sets. Nucleic Acids Res. (2021) 49:D605–d612. doi: 10.1093/nar/gkaa1074
21
YounessMEkhteraei-TousiSNagarajuCKPuertasRDThienpontBRegaFet al. Location and aetiology are determinants of fibroblast activation and heterogeneity in the failing human heart. Genome Med. (2025) 17:155. doi: 10.1186/s13073-025-01580-z
22
LengDYangZSunHSongCHuangCIpKUet al. Comprehensive analysis of tumor microenvironment reveals prognostic ceRNA network related to immune infiltration in sarcoma. Clin Cancer Res. (2023) 29:3986–4001. doi: 10.1158/1078-0432.ccr-22-3396
23
ZengDYeZShenRYuGWuJXiongYet al. IOBR: multi-omics immuno-oncology biological research to decode tumor microenvironment and signatures. Front Immunol. (2021) 12:687975. doi: 10.3389/fimmu.2021.687975
24
FinotelloFMayerCPlattnerCLaschoberGRiederDHacklHet al. Molecular and pharmacological modulators of the tumor immune contexture revealed by deconvolution of RNA-seq data. Genome Med. (2019) 11:34. doi: 10.1186/s13073-019-0638-6
25
HaoYStuartTKowalskiMHChoudharySHoffmanPHartmanAet al. Dictionary learning for integrative, multimodal and scalable single-cell analysis. Nat Biotechnol. (2024) 42:293–304. doi: 10.1038/s41587-023-01767-y
26
ChenYYimWYLiCWangZHouJPengYet al. T cell senescence induced by DNMT1 deletion mediates cardiac transplant tolerance via altered CDKN1A methylation. Mol Ther. (2026) 34(6):3484–502. doi: 10.1016/j.ymthe.2026.02.015
27
HuCLiTXuYZhangXLiFBaiJet al. CellMarker 2.0: an updated database of manually curated cell markers in human/mouse and web tools based on scRNA-seq data. Nucleic Acids Res. (2023) 51:D870–d876. doi: 10.1093/nar/gkac947
28
LiKDuYCaiYLiuWLvYHuangBet al. Single-cell analysis reveals the chemotherapy-induced cellular reprogramming and novel therapeutic targets in relapsed/refractory acute myeloid leukemia. Leukemia. (2023) 37:308–25. doi: 10.1038/s41375-022-01789-6
29
ForsbergMKonoplevaM. AML treatment: conventional chemotherapy and emerging novel agents. Trends Pharmacol Sci. (2024) 45:430–48. doi: 10.1016/j.tips.2024.03.005
30
LinHWangTWeiJ. Strategic innovations: Tackling challenges of immunotherapy in acute myeloid leukemia. Chin J Cancer Res. (2025) 37:490. doi: 10.21147/j.issn.1000-9604.2025.04.02
31
MayerhoferCLiDKristiansenTMayerhoferEShardaASchiroliGet al. OGFOD1 enables AML chemo- and nutrient stress resistance by regulating protein synthesis. Cell Metab. (2025) 37:2014–29:e2019. doi: 10.1055/s-0046-1822682
32
GuoHZFengRXZhangYJYuYHLuWLiuJJet al. A CD36-dependent non-canonical lipid metabolism program promotes immune escape and resistance to hypomethylating agent therapy in AML. Cell Rep Med. (2024) 5:101592. doi: 10.1016/j.xcrm.2024.101592
33
Pakos-ZebruckaKKorygaIMnichKLjujicMSamaliAGormanAM. The integrated stress response. EMBO Rep. (2016) 17:1374–95. doi: 10.15252/embr.201642195
34
Costa-MattioliMWalterP. The integrated stress response: From mechanism to disease. Science. (2020) 368(6489):eaat5314. doi: 10.1126/science.aat5314
35
PettiAAKhanSMXuZHeltonNFronickCCFultonRet al. Genetic and transcriptional contributions to relapse in normal karyotype acute myeloid leukemia. Blood Cancer Discov. (2022) 3:32–49. doi: 10.1158/2643-3230.bcd-21-0050
36
SeverensJFKarakaslarEOvan der ReijdenBASánchez-LópezEvan den BergRRHalkesCJMet al. Mapping AML heterogeneity - multi-cohort transcriptomic analysis identifies novel clusters and divergent ex-vivo drug responses. Leukemia. (2024) 38:751–61. doi: 10.1038/s41375-024-02137-6
37
ZhouFArouaNLiuYRohdeCChengJWirthAKet al. A dynamic rRNA ribomethylome drives stemness in acute myeloid leukemia. Cancer Discov. (2023) 13:332–47. doi: 10.1158/2159-8290.cd-22-0210
38
MaurerKParkCYManiSBorjiMRathsFGouinKHet al. Coordinated immune networks in leukemia bone marrow microenvironments distinguish response to cellular therapy. Sci Immunol. (2025) 10:eadr0782. doi: 10.1126/sciimmunol.adr0782
39
RadpourRSimillionCWangBAbbasHARietherCOchsenbeinAF. IL-9 secreted by leukemia stem cells induces Th1-skewed CD4+ T cells, which promote their expansion. Blood. (2024) 144:888–903. doi: 10.1182/blood.2024024000
40
VoltaVPérez-BaosSde la ParraCKatsaraOErnlundADornbaumSet al. A DAP5/eIF3d alternate mRNA translation mechanism promotes differentiation and immune suppression by human regulatory T cells. Nat Commun. (2021) 12:6979. doi: 10.1038/s41467-021-27087-w
Summary
Keywords
acute myeloid leukemia, EIF2S1, immune micro-environment, metabolic remodeling, prognostic biomarker, single-cell RNA sequencing
Citation
Hong X, Huang Y, Wu W, Jiang X, Lin Y, Xue Y and Lin D (2026) Dual relevance of EIF2S1 in acute myeloid leukemia: linking poor prognosis to immune-metabolic states and leukemic cell fitness. Front. Oncol. 16:1922910. doi: 10.3389/fonc.2026.1922910
Received
29 June 2026
Revised
18 July 2026
Accepted
21 July 2026
Published
05 August 2026
Volume
16 - 2026
Edited by
Yonghui Li, Shenzhen University General Hospital, China
Reviewed by
Chong Zhang, Southwest University, China
Rajiv Ranjan Kumar, National Cancer Institute, AIIMS, India
Updates
Copyright
© 2026 Hong, Huang, Wu, Jiang, Lin, Xue and Lin.
This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.
*Correspondence: Donghong Lin, lindh65@fjmu.edu.cn; Yan Xue, 9200501038@fjmu.edu.cn
†These authors have contributed equally to this work
Disclaimer
All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article or claim that may be made by its manufacturer is not guaranteed or endorsed by the publisher.