Fatty acid desaturase 1 drives esophageal cancer via M1 macrophage ferroptosis and lipid metabolism remodeling

Abstract

BACKGROUND

Esophageal cancer is a major health challenge with a poor prognosis. The tumor microenvironment, particularly the role of immune cells such as macrophages, is critical in cancer progression. The lipid metabolism enzyme fatty acid desaturase 1 (FADS1) is involved in various cancers, but its specific function in the immune regulation of esophageal cancer is unclear.

AIM

To explore whether FADS1 promotes esophageal cancer progression by remodeling the phospholipid metabolism, facilitating lipid raft assembly and triggering ferroptosis in M1 macrophages.

METHODS

We conducted a bidirectional Mendelian randomization study to identify ferroptosis-related genes linked to esophageal cancer susceptibility. Comprehensive bioinformatics analyses were performed using public transcriptomic and single-cell RNA sequencing datasets. Protein localization was examined via immunofluorescence in clinical specimens. Laboratory experiments involved the genetic manipulation of Tamm-Horsfall protein-1-derived M1 macrophages, measurement of lipid oxidation products, coculture assays, phospholipid supplementation, and disruption of lipid rafts. Virtual gene knockout and artificial intelligence-guided drug screening were also employed.

RESULTS

FADS1 was identified as a causal factor for esophageal cancer development and was overexpressed in tumor tissues, particularly in M1-polarized macrophages. Genetic inhibition of FADS1 expression in M1 macrophages reduced lipid peroxidation and ferroptosis, enhancing their cancer-killing ability in coculture with esophageal cancer cells. Mechanistically, FADS1 promoted the accumulation of 1-palmitoyl-2-stearoyl-glycerophosphocholine (16:0/18:0), which facilitated lipid raft assembly and triggered ferroptosis in M1 macrophages. Computational analysis revealed an immune-related gene network controlled by FADS1. The compound BRD-K95779464 was identified as a promising FADS1 inhibitor.

CONCLUSION

FADS1 promotes esophageal cancer progression by reprogramming the phospholipid metabolism to assemble lipid rafts, which triggers ferroptosis in M1 macrophages. This study identifies FADS1 as a potential therapeutic target for esophageal cancer immunotherapy.

Key Words: Fatty acid desaturase 1; Esophageal cancer; Ferroptosis; Phospholipid metabolism; Lipid raft

Core Tip: Esophageal cancer patients have a poor prognosis with unclear immunometabolic mechanisms. Fatty acid desaturase 1 (FADS1), a key lipid metabolism enzyme, is linked to multiple cancers, but its role in esophageal cancer immunity is unknown. In this study, FADS1, which is overexpressed in tumor-infiltrating M1 macrophages, was identified as a causal factor for esophageal cancer. Mechanistically, FADS1 accumulates 1-palmitoyl-2-stearoyl-glycerophosphocholine, promotes lipid raft assembly, and triggers ferroptosis in M1 macrophages, impairing their antitumor function. Inhibiting FADS1 expression suppressed ferroptosis and restored M1 macrophage activity. The FADS1 inhibitor BRD-K95779464 was identified by artificial intelligence screening. This work reveals a novel FADS1-mediated immunometabolic pathway in esophageal cancer.

INTRODUCTION

Esophageal cancer is a major worldwide public health problem since its occurrence and mortality rates are persistently high[1]. The prognosis is poor for most patients, even with advances in detection methods and therapeutic intervention, because of the aggressiveness of this type of malignancy and because it is usually identified when the disease has reached a late phase[2]. Currently, most ongoing studies emphasize the role of the tumor microenvironment in oncogenesis, particularly the complex molecular dialog of cancerous cells with immune elements, which, in general, affects pathological evolution[3]. Among immunological components, macrophages are a dominant cell type that is highly plastic and can adopt different functional states that can participate in suppressing or facilitating tumor growth[4]. This functional diversity highlights the complex role of macrophages in esophageal carcinogenesis and highlights the need for a thorough exploration of the underlying molecular processes that regulate their observable trait variations.

The regulation of lipid metabolism is fundamental to cell functions and substantially influences tumorigenesis and disease progression[5]. Fatty acid desaturase 1 (FADS1) is an important enzyme that governs the so-called 5-desaturation step in polyunsaturated fatty acid biosynthesis; therefore, it is an important subject in oncological metabolism research[6]. It is an enzyme that allows the conversion of dietary linoleic acid to arachidonic acid[7]. This metabolism affects the composition of membranes and cellular signaling but also influences the inflammatory response and apoptosis regulation. There is scientific evidence that many cancers exhibit abnormal FADS1 expression, which affects neoplastic proliferation and migration through changes in lipid metabolism and downstream signaling cascades[8-10]. However, a detailed examination of the specific functions of FADS1 in immune cell populations, particularly the macrophages in the tumor microenvironment, is lacking. The complex immunological microenvironment that is present in tumors limits the complete understanding of its possible role in the progression of esophageal cancer.

Ferroptosis is a recently reported form of programmed cell death characterized by iron-dependent lipid peroxidation, and its role in tumor immune regulation is being increasingly recognized[11]. M1 macrophages have pronounced antitumor properties because they are proinflammatory macrophages, and the functional state of M1 macrophages is highly correlated with ferroptosis pathways[12,13]. Recent studies have demonstrated that ferroptosis in macrophages can be targeted and modulated to change both the phenotypic polarization and the secretion of cytokines and subsequently influence immune responses in tumor microenvironments[14,15]. However, the specific mechanism through which FADS1 may regulate the ferroptosis of M1 macrophages through lipid metabolism and its effects on the development of esophageal carcinoma are unexplored and urgently need to be studied.

The present study focused on analyzing the role of FADS1 in the tumor microenvironment of esophageal cancer and, in particular, its role in ferroptosis mechanisms in M1 macrophages. This work contributes to new areas of research that involve lipid metabolism and programmed cellular demise, and it explores how FADS1 regulates the restructuring of phospholipid metabolism to influence both the dynamics of lipid peroxidation and ferroptosis-related signaling cascades. These biological processes in turn change the functionality of macrophages and their ability to modulate antitumor immune responses in esophageal malignancies. This work fills a relevant gap in knowledge on how FADS1 is connected to macrophage ferroptosis and reveals a possible new target for immunotherapeutic intervention in esophageal carcinoma, providing interesting suggestions for both fundamental research and clinical practice.

MATERIALS AND METHODS

Study design and data sources

Data sources for bioinformatics and genetic analyses: The gene set associated with ferroptosis used in this work is from the Ferrdb repository (http://www.zhounan.org/ferrdb/current/). Information on the cis-expression quantitative trait loci for some genes was obtained from the eQTLGen platform (https://www.eqtlgen.org/). The information from the genome-wide association study for esophageal carcinoma is from the OpenGWAS resource (ID: Bbj-a-117; https://opengwas.io/). The transcriptomic information of the esophageal cancer specimens is from the GSE273848 dataset in the Gene Expression Omnibus repository[16], and the single-cell RNA sequencing information is from the GSE196756 dataset. The summary statistics from the metabolite genome-wide association study analyses are from previous large-scale cohort investigations[17]. The protein structure information is from the UniProt database (https://www.uniprot.org/), and the configurations of the small molecules are from PubChem (https://pubchem.ncbi.nlm.nih.gov/).

Experimental samples and cells: Tissue samples from esophageal squamous cell carcinoma and corresponding healthy adjacent tissues from the same patients, fixed in paraffin blocks for immunofluorescence, were obtained from Shanghai Xuhui District Central Hospital. Among the three patients, two were male, and one was female; these patients were aged 71, 72, and 67 years. According to the 8^th edition of the American Joint Committee on Cancer Tumor-Node-Metastasis staging system for esophageal cancer, their pathological tumor stages were IIIB, IIA, and IB, respectively. Tohoku Hospital Pediatrics-1 (THP-1) cells (RRID: CVCL_0006; iCell-h213; iCell Bioscience, Shanghai, China) were used as a model system to study macrophage differentiation and the functional characteristics of macrophages under laboratory conditions, while OE-19 cells (RRID: CVCL_1622; iCell-h391; iCell Bioscience, Shanghai, China) were used to test the growth of malignant esophageal cells. All the cell lines were authenticated by short-tandem-repeat profiling within the past three years and are available in the ExPASy Cellosaurus database under the indicated RRIDs.

Mendelian randomization analysis

In the first phase, a two-sample Mendelian randomization (MR) study was performed, in which the metabolites were the exposures and esophageal cancer was the outcome. We chose genetic variants that passed very strict criteria, namely, single nucleotide polymorphisms with P values smaller than 5 ×10^-8 (i.e., genome-wide significance), those without links (linkage disequilibrium r² smaller than 0.001), those with a long physical distance (greater than 10000 kb between the variants), and those with high instrument strength (F-statistic greater than 10). The inverse variance weighted technique was the main statistical technique used for this analysis. A separate MR investigation was subsequently performed with the cis-expression quantitative trait loci of the target genes as the exposures and the levels of metabolites as the outcomes. The mediation effect was calculated as the product of the coefficients (β₁ × β₂), and the standard error was calculated by the delta method formula: SE = β12​×SE22 ​+β​22×SE12. The proportion of mediation was calculated as the indirect effect divided by the total observed effect.

Transcriptomic differential expression analysis

The GSE273848 dataset was normalized, and some preprocessing procedures were performed with the R programming environment (v4.5.2) and the limma toolkit. Missing data were filled through implementation of the k-nearest neighbors approach, and the redundant probes were combined with the avereps functionality. Specific gene expression profiles [FADS1, FADS2, long-chain fatty acid elongase 5 (ELOVL5), SRC proto-oncogene (SRC), and peroxisome proliferator-activated receptor gamma (PPARG)] were obtained. The study population was divided into two cohorts: Esophageal carcinoma patients (22) and healthy controls (18). Comparisons between groups were performed with the Wilcoxon nonparametric test.

Single-cell RNA sequencing data analysis

Data preprocessing and cell clustering: We used Seurat software (v5.4.0) for quality assessment. We set the following thresholds: 200-7500 genes per cell were detected, the minimum number of UMI counts was greater than 1000, and the mitochondrial gene content was less than 25%. We used the decontX method to remove the background of RNA contamination, DoubletFinder to remove potential doublet cells, SCTransform to normalize the data, and the harmony approach for batch effect correction. For dimensional reduction, we chose the first 20 principal components for downstream analysis. Cell populations were identified by means of graph-based clustering (set to 0.3), and dimensional reduction was used to visualize the results by t-SNE and uniform manifold approximation and projection (UMAP) techniques. The class of the cell type was determined by using SingleR with the human primary cell atlas as a reference, as well as manual checking according to known marker genes.

Differential expression and functional enrichment: Differentially expressed genes were identified by the FindAllMarkers function using the Wilcoxon rank-sum test, with selection on the absolute log₂ fold change greater than 1 and a Bonferroni corrected significance level less than 0.05. The functional annotation of the differentially expressed genes was performed by Gene Ontology analysis with the clusterProfiler package, by a hypergeometric test and with a threshold of P < 0.05. Macrophage subpopulations were also removed and subjected to further clustering procedures to determine their diversity. The transcription factor regulatory potential was estimated by pySCENIC. Ferroptosis-related gene expression patterns were quantified by single-sample gene set enrichment analysis, implemented in the GSVA package.

Cell-cell communication network analysis: Intercellular communication networks were constructed with CellChat software (version 1.6.1) with the standard configuration settings. The ligand-receptor interactions were significantly related to the communication intensity and the signaling pathway activation for the different cell populations according to permutation testing (100 times, P < 0.05).

Ferroptosis signature scoring and diagnostic model construction: We used the AddModuleScore function of the Seurat package to create a score for a single-cell diagnostic model; that is, we used the five principal ferroptosis-associated genes (FADS1, FADS2, ELOVL5, SRC, and PPARG) that we identified by MR analysis. This measure is the standard enrichment level when the mean expression of a specific gene is compared with that of a random gene collection. Ferroptosis activity levels were also evaluated by ssGSEA.

scTenifoldKnk virtual gene knockout analysis: FADS1 was selectively inactivated by computational simulation with the scTenifoldKnk toolkit under the following operational criterion: A cutoff of 0.1 for the content of the mitochondrial genes, with 1000 sequencing reads per cell as the basal, 2000 most variable genes for network modeling, and 3 different network modules and 300 cells per module chosen at random. Statistical evaluation was performed by computing the Z scores to estimate the changes in regulation and by applying multiple testing correction to obtain the adjusted probability values (P_adj) by comparing the architectures of the gene networks before and after intervention. Genes whose expression significantly changed after the suppression of FADS1 expression were identified on the basis of a corrected significance threshold P_adj < 0.05.

Experimental validation methods

Tissue immunofluorescence staining: The tissue samples of esophageal carcinoma and the surrounding healthy regions, in paraffin blocks, were first treated with xylene (to remove paraffin under coating) and then rehydrated with an ethanol gradient series. The samples were subjected to antigen exposure by high-temperature treatment in a citrate (pH 6.0)-based solution under pressure for two minutes. The tissue sections were treated with 0.5% Triton X-100 solution to increase permeability and then exposed to goat serum for nonspecific binding. The prepared sections were treated with a primary antibody containing rabbit-derived anti-FADS1 antibodies (Proteintech, 10627-1-AP; diluted 1:300, IL, United States) and incubated at 37 °C for one hour. After thorough rinsing with phosphate-buffered saline, the samples were exposed to the secondary antibody, fluorescein isothiocyanate-labeled goat anti-rabbit immunoglobulin (ebiogo, No. B029, Lot No. 09302309, China). The samples were incubated with a 1:500 dilution of the primary antibody at 37 °C for half an hour in the dark. The tissue sections were coverslipped with 4’,6-diamidino-2-phenylindole-containing antifade mounting solution. Digital imaging was performed with a Pannoramic MIDI slide scanner from 3DHISTECH. Quantitative analysis was performed with ImageJ software to determine the integrated optical density (OD) and the positively stained regions. The fluorescence intensity values were thus obtained by calculating the ratio of the integrated OD to the measured area.

Cell culture, differentiation, and transfection: For the cellular studies, the procedures were as follows: Cell line maintenance, induction of differentiation, and genetic manipulation (transfection). The THP-1 cells were differentiated into macrophage phenotypes by treatment with 100 ng/mL phorbol-12-myristate-13-acetate for 2 days. These M1 macrophages were subsequently polarized by treatment with 100 ng/mL lipopolysaccharide and 20 ng/mL interferon-γ for 48 hours. FADS1 gene silencing was performed by transfection with three different siRNA constructs (si-FADS1-1, si-FADS1-2, and si-FADS1-3) and a nontargeting control with the reagent lipofectamine 3000 for 48 hours. The efficiency of gene knockdown was evaluated by quantitative polymerase chain reaction analysis. The oligonucleotide sequences used for FADS1 detection were as follows: Forward primer 5’-CCAACTGCTTCCGCAAAGAC-3’ and reverse primer 5’-GCTGGTGGTTGTACGGCATA-3’; for the reference gene, β-actin: Forward primer 5’-AGCACAGAGCCTCGCCTTTG-3’ and reverse primer 5’-CTTCTGACCCATGCCCACCA-3’.

Malondialdehyde content assay: After treatment, M1 macrophages were collected and subjected to several freeze-thaw cycles to lyse the cells. The lysates were subsequently centrifuged to obtain the supernatant fractions. Malondialdehyde (MDA) levels were measured with a commercially available human MDA enzyme-linked immunosorbent assay kit (Coibo Biological, CB10401-Hu, Shanghai, China) according to the supplier’s instructions. The OD was read at a wavelength of 450 nm with a microplate spectrophotometer (Thermo Fisher, FC, MA, United States). The MDA concentrations were calculated by comparison with the calibration curve.

Lipid peroxidation assay (C11-BODIPY 581/591 staining): To test for lipid peroxidation, treated M1 macrophages were incubated with 2 μM BODIPY™ 581/591 C11 fluorescent dye (BODIPY; Beyotime, S0043S, Shanghai, China), prepared by diluting a 2 mM stock solution 1:1000 in PBS, for 10 minutes at 37 °C under light protection. Nuclear staining was performed with Hoechst 33342. Fluorescence microscopic analysis (Olympus, CKX53, Tokyo, Japan) was performed by observing two different emission spectra, where oxidized probes emitted green fluorescence and reduced probes red fluorescence.

Cell viability assay and coculture experiment: To observe the interaction between M1 macrophages and esophageal carcinoma cells (OE-19), we established a coculture system in a Transwell chamber. In the experiment, OE-19 cells were placed in the bottom compartment, and M1 macrophages (with siRNA negative control or si-FADS1 modification) were placed in the upper compartment. Cell viability was measured at different times (0, 24, 48, 72 and 96 hours) by cell counting kit-8 reagent from Beyotime, and the OD was measured at 450 nm after a 60-minute incubation.

Phospholipid rescue and lipid raft inhibition assays: To test whether FADS1 is involved in the regulation of ferroptosis through these particular phospholipid pathways, two concentrations (10 μmol/L and 32 μmol/L) of 1-palmitoyl-2-stearoyl-glycerophosphocholine (GPC) (16:0/18:0) were added to the cells in which the expression of FADS1 was suppressed. To study lipid raft functionality, the cellular samples were treated for 24 hours with 1 mmol/L methyl-β-cyclodextrin (HY-101461 from MCE, NJ, United States) to induce disintegration of the structure of the lipid rafts.

Lipid raft marker immunofluorescence: The cellular specimens were fixed with 4% paraformaldehyde and treated with 0.1% Triton X-100 for membrane permeabilization. The cellular samples were incubated with rabbit-derived anti-caveolin-1 (Proteintech, 16447-1-AP; 1:250 dilution, IL, United States) as the primary antibody at 4 °C for approximately 16 hours. The cellular samples were subsequently incubated with Alexa Fluor 488-labeled secondary antibodies (Abcam, ab150077; 1:500 dilution, MA, United States) under ambient conditions for 60 minutes. After nuclear staining with 4’,6-diamidino-2-phenylindole, cellular images were observed by fluorescence microscopy.

Artificial intelligence-assisted drug screening and molecular docking

Candidate drug screening: The gene-specific v-scores (representing effect sizes) were calculated from the GSE273848 dataset by using the equation: (x¯tumor-x¯normal)/σtumor2+σnormal2. The genome-wide v-score distributions were then run through the pretrained DrugReflector algorithm to obtain the reversal scores (as logit probabilities) for the potential therapeutic compounds.

Molecular docking: Molecular docking simulations were carried out using the CB-Dock2 web-based interface (http://cadd.labshare.cn/cb-dock2/) and the computational algorithm AutoDock Vina. The docking protocol involved semiflexible parameters, keeping the structure of the protein as is and the ligand flexible. The binding interaction was considered significant if the Vina energy score was -5.0 kcal/mol or lower.

Statistical analysis

All the statistical computations were performed using the R programming environment (release 4.5.2) or IBM SPSS Statistics 26. Quantitative results are given as the arithmetic mean ± SD. For two-group comparisons, independent samples t tests with two-tailed distributions were used. For multiple group analysis, one-way or two-way analysis of variance was used, and pairwise comparisons were performed using Tukey’s honestly significant difference test. In the assessment of computational biology, for the analysis of continuous variables, the nonparametric Wilcoxon rank-sum test (for two groups) or the Kruskal-Wallis test (for multiple groups) was performed. For investigations of MR, the inverse-variance weighted technique was mostly used, and some other robustness checks involving sensitivity analyses were performed. The P value threshold for statistical significance was P < 0.05.

RESULTS

MR identifies causal associations of five ferroptosis-related genes, including FADS1, with esophageal cancer risk

After the instrumental variables were selected, the 483 genes related to ferroptosis were subjected to two-sample MR analysis. In the main analysis using inverse-variance weighted analysis, the genetic variants of five genes were significantly associated with susceptibility to esophageal cancer (P < 0.05; Table 1), including FADS1 [odds ratio (OR) = 0.75; 95% confidence interval (CI): 0.62-0.91; P = 0.0038], FADS2 (OR = 0.84; 95%CI: 0.76-0.93; P = 0.0007), SRC (SRC proto-oncogene; OR = 0.11; 95%CI: 0.02-0.71; P = 0.0206), ELOVL5 (OR = 1.63; 95%CI: 1.03-2.60; P = 0.0384), and PPARG (OR = 0.62; 95%CI: 0.40-0.99; P = 0.0445). Other sensitivity assessments, e.g., weighted median and MR-Egger regression analyses, confirmed these results and revealed no serious evidence of horizontal pleiotropy or heterogeneity. These MR results reflect only population-level genetic susceptibility and do not represent the pathological expression or function of FADS1 in local tumor lesions; subsequent tissue and single-cell sequencing will further elucidate the local expression characteristics of FADS1.

Table 1 Mendelian randomization estimates for the causal associations of five ferroptosis-related genes with esophageal cancer risk using the inverse variance weighted method.

ID	NSNP	Method	β	SE	OR	OR_lci95	OR_uci95	P value
FADS2	14	IVW	-0.172	0.051	0.842	0.763	0.930	0.0007
FADS1	9	IVW	-0.282	0.097	0.755	0.624	0.913	0.0038
SRC	3	IVW	-2.202	0.951	0.111	0.017	0.713	0.0206
ELOVL5	3	IVW	0.492	0.237	1.635	1.027	2.603	0.0384
PPARG	7	IVW	-0.470	0.234	0.625	0.395	0.989	0.0445

FADS2: Fatty acid desaturase 2; FADS1: Fatty acid desaturase 1; SRC: SRC proto-oncogene, non-receptor tyrosine kinase; ELOVL5: Elongation of very long chain fatty acids protein 5; PPARG: Peroxisome proliferator-activated receptor gamma; IVW: Inverse variance weighted; NSNP: Number of single nucleotide polymorphisms; OR_lci95: Odds ratio lower 95% confidence interval; OR_uci95: Odds ratio upper 95% confidence interval.

Key lipid metabolism genes, including FADS1, are upregulated in esophageal cancer tissues

According to the bioinformatics data of 22 esophageal tumor samples and 18 normal esophageal tissue samples, the FADS1 expression level in cancerous tissue was markedly higher than that in healthy tissue (Figure 1A). Similar observations were also made for the FADS2, PPARG, and SRC genes. All the mentioned genes were strongly overexpressed in malignant tissues (Figure 1B-D). However, no significant differential expression of ELOVL5 was detected between the two sample groups (Figure 1E). Thus, the results above suggest that genes involved in lipid metabolism, especially FADS1 and FADS2, may play some role in esophageal carcinogenesis.

Figure 1 Expression levels of 5 genes in esophageal cancer tissues. A: The expression level of fatty acid desaturase 1 in esophageal cancer tissues was significantly greater than that in normal esophageal tissue; B: The expression level of fatty acid desaturase 2 in esophageal cancer tissues was significantly greater than that in normal tissues; C: The expression level of elongation of very long chain fatty acids protein 5 in esophageal cancer tissues was significantly greater than that in normal tissues; D: The expression level of SRC proto-oncogene in esophageal cancer tissues was significantly greater than that in normal tissues; E: No significant difference in peroxisome proliferator-activated receptor gamma expression was detected between esophageal cancer and normal tissues. ^aP < 0.05, ^cP < 0.0001, NS: Not significant (P ≥ 0.05). FADS1: Fatty acid desaturase 1; FADS2: Fatty acid desaturase 2; ELOVL5: Elongation of very long chain fatty acids protein 5; SRC: SRC proto-oncogene (sarcoma gene); PPARG: Peroxisome proliferator-activated receptor gamma.

Mediating effects of metabolites on the association between FADS1 genetic variants and esophageal cancer risk

The results of the above two-stage MR mediation analysis revealed that the genetic variants in FADS1 were indeed significantly associated with the concentrations of different kinds of lipid metabolites and that the concentrations of some of these metabolites were significantly associated with susceptibility to esophageal cancer. The results of the mediation analysis revealed that the genetic effect of FADS1 on the development of esophageal cancer was partly transmitted by the concentrations of some particular phospholipid metabolites. For example, 1-stearoyl-2-arachidonoyl-glycerophosphoethanolamine (18:0/20:4) had a mediating effect size of -0.116, which explained 41.0% of the total effect (P = 0.00098), and 1-linoleoyl-2-arachidonoyl-GPC (18:2/20:4n-6) explained 47.6% (P = 0.00081) (Table 2). These results indicated that FADS1 might modulate susceptibility to esophageal cancer by affecting the metabolic profiles of some particular phospholipid compounds, among which those containing arachidonic acid were of particular interest.

Table 2 Key mediator metabolites identified in the fatty acid desaturase 1-mediated metabolic pathway analysis.

Metabolite	β1	β2	Mediation effect (β1 × β2)	Mediation proportion	P value
1-linoleoyl-GPE (18:2)	0.446	-0.285	-0.127	45.12%	0.006385603
1,2-dilinoleoyl-GPC (18:2/18:2)	0.451	-0.280	-0.126	44.84%	0.006059174
1-palmitoyl-2-stearoyl-gpc (16:0/18:0)	-0.399	0.449	-0.179	63.64%	0.006666497
1-stearoyl-2-arachidonoyl-GPE (18:0/20:4)	-0.392	0.294	-0.116	41.04%	0.000979761
1-linoleoyl-2-arachidonoyl-GPC (18:2/20:4n-6)	-0.559	0.240	-0.134	47.61%	0.000806059

β₁: Effect of fatty acid desaturase 1 on metabolite (inverse variance weighted); β₂: Effect of metabolite on esophageal cancer risk (inverse variance weighted); mediation effect = β₁ × β₂; mediation proportion = (mediation effect/total effect) × 100%; P value from the delta method (two-sided); GPE: Glycerophosphoethanolamine; GPC: Glycerophosphocholine.

Single-cell transcriptomic atlas reveals the cellular heterogeneity of the esophageal cancer microenvironment and myeloid cell-enriched expression of FADS1

After strict quality assessment and the full process of GSE196756, 24196 single-cell transcriptomic profiles with strict quality were obtained. By using the t-SNE and UMAP techniques for dimension reduction and after clustering, the following seven main cellular populations were obtained: T lymphocytes (39.01%), B lymphocytes (21.90%) and monocyte-derived macrophages (8.93%), as well as neutrophils, mast cells, fibroblasts and vascular smooth muscle cells (Figure 2A-C). We further analyzed the macrophage subsets and found that FADS1 was preferentially expressed in M1-polarized macrophages. We also found that the ferroptosis activity score was much greater for M1 macrophages than for M2 macrophages (Figure 2D). These findings suggest that FADS1 may play a role in modulating the tumor microenvironment through ferroptosis in M1 macrophages. An evaluation of the single-cell expression profiles for the MR-identified critical genes (FADS1, FADS2, ELOVL5, SRC, and PPARG) revealed enrichment of myeloid cells for FADS1, particularly for the monocyte-macrophage lineages and neutrophil populations. The results of the quantitative evaluation, visualized as a violin plot, confirmed that the transcriptional levels of FADS1 were significantly increased in monocytic cells. We observed that the expression levels were also greater in monocyte/macrophage populations than in lymphocytes and in other cellular populations (statistical analysis, Kruskal-Wallis test; P < 0.001) (Figure 2E). These observations prompted subsequent studies of the mechanisms involved only in tumor-infiltrating myeloid cell populations.

Figure 2 Single-cell transcriptomic landscape of the esophageal cancer microenvironment and distribution of fatty acid desaturase 1-associated signatures. A: Bar plot of cell type proportions across samples; B: Uniform manifold approximation and projection (UMAP) plot illustrating the two-dimensional distribution characteristics of each cell type; C: UMAP plot of the 7 major cell types; D: Subclustering of macrophages showing fatty acid desaturase 1 enrichment in M1-like macrophages and ferroptosis scores in M1 vs M2 macrophages. Ferroptosis scores were calculated using single-sample gene set enrichment analysis; E: Violin plot showing fatty acid desaturase 1 expression across different cell types; F: UMAP plot of diagnostic model scores; G: Violin plot comparing diagnostic model scores across cell types; H: CellChat cell-cell communication network plot showing the number of ligand-receptor interactions among different cell types; I: CellChat network plot showing the strength (weight) of signaling interactions among cell types; J: Ligand-receptor network of the macrophage migration inhibitory factor and galactoside binding lectin signaling pathways. UMAP: Uniform manifold approximation and projection; CMP: Common myeloid progenitor; SMC: Smooth-muscle cell; TSC: Tumor stem cell; FADS1: Fatty acid desaturase 1; HSC: Hematopoietic stem cell.

Ferroptosis-related diagnostic model scores are significantly elevated in monocytes/macrophages

UMAP visualization of the scoring system of the diagnostic model obtained from the five key genes of the cellular distribution is shown. The cells with high scores were located mainly in myeloid populations, while in the T-cell and B-cell zones, the values were slightly lower (Figure 2F). By quantitative statistical analysis, we observed that monocytes, common myeloid progenitors and tissue stem cells had the most significant model scores, which were much higher than those of the lymphocyte populations (Kruskal-Wallis test, P < 0.001) (Figure 2G). The spatial distribution of the scores is the same as that of the high expression profile of FADS1 and related genes, indicating that the ferroptosis-associated signaling pathway identified by this diagnostic framework is focused mainly on monocyte/macrophage lineages.

Cell-cell communication network analysis uncovers the core regulatory role of myeloid cells in the microenvironment

An investigation of the intercellular communication using CellChat revealed complex ligand-receptor interactions in the esophageal tumor microenvironment. Among all the cell types, fibroblasts and monocytes have the most extensive and most robust connectivity patterns and are therefore the most important nodes in the cellular signaling network (Figure 2H and I). A detailed examination of the different signaling cascades revealed that the macrophage migration inhibitory factor signaling route is very prominent in mediating the crosstalk between myeloid-derived cells and lymphocyte populations. On the other hand, the galactoside binding lectin signaling axis is particularly strong and is driven mainly by the binding of lectin, galactoside-binding, soluble 9 molecules secreted from myeloid cells to hepatitis A virus cellular receptor 2 immune checkpoint molecules expressed on T lymphocytes. The molecular interaction between lectin, galactoside-binding, soluble 9 and hepatitis A virus cellular receptor 2 may be involved in promoting T-cell dysfunction and facilitating tumor immune evasion (Figure 2J).

FADS1 protein is upregulated in esophageal cancer tissues and mainly localized in the cytoplasm

To verify the above bioinformatics prediction, immunofluorescence analysis was performed on cancerous and adjacent noncancerous esophageal tissues. Compared with that in normal tissues, FADS1 in malignant tissues was predominantly localized in the cytoplasm, and the signal intensity in malignant tissues was markedly increased (Figure 3A). Compared with that in normal controls, the average fluorescence density (integrated OD/area) in tumor samples was significantly greater (0.432 vs 0.207; P < 0.001) (Figure 3B), providing further evidence for the role of FADS1 in esophageal carcinogenesis.

Figure 3 Fatty acid desaturase 1 is upregulated in esophageal cancer tissues. A: Representative immunofluorescence images of fatty acid desaturase 1 (FADS1) in esophageal cancer and adjacent normal tissues. FADS1 is shown in green, and nuclei are stained with 4’,6-diamidino-2-phenylindole; B: Quantitative analysis of the mean fluorescence intensity (integrated optical density/area) of FADS1 in normal and tumor tissues. P < 0.001 vs the normal group. FADS1: Fatty acid desaturase 1; DAPI: 4’,6-diamidino-2-phenylindole.

Validation of FADS1 knockdown efficiency and its effect on ferroptosis in M1-type macrophages

To address the role of FADS1 in macrophage biology, siRNA-mediated gene silencing was used in M1-polarized macrophages. The siRNA construct that had the best FADS1 silencing efficacy was si-FADS1-1, with a 53.2% decrease in FADS1 mRNA levels compared with those in the control (P < 0.001) (Figure 4A and B). Biochemically, the MDA concentration strongly decreased after FADS1 suppression (0.278 ± 0.006 nmol/mL vs 0.464 ± 0.021 nmol/mL in the control, P < 0.001) (Figure 4C), indicating decreased lipid peroxidation activity. These findings indicate that FADS1 participates in ferroptosis processes in M1 macrophages. The analytical validity of the MDA measurements is shown by a calibration curve with near-perfect linearity (r² = 0.9997) (Figure 4D). Similar evidence was obtained by C11-BODIPY 581/591 staining, in which FADS1 depletion resulted in an attenuated oxidative signal (green fluorescence) and an intensified reduced-state signal (red fluorescence). The data shown in Figure 5 reveal a decrease in lipid peroxidation and show that FADS1 promotes ferroptosis in M1-polarized macrophages.

Figure 4 Fatty acid desaturase 1 knockdown reduces malondialdehyde levels and lipid peroxidation in M1 macrophages. A: Western blot analysis of fatty acid desaturase 1 (FADS1) protein expression after knockdown; B: Relative FADS1 mRNA expression in macrophages transfected with siRNA negative control or si-FADS1; C: Malondialdehyde levels in M1 macrophages after FADS1 knockdown; D: Standard curve for malondialdehyde quantification. P < 0.001 vs siRNA negative control. FADS1: Fatty acid desaturase 1; MDA: Malondialdehyde; siNC: SiRNA negative control.

Figure 5 Fatty acid desaturase 1 knockdown attenuates lipid peroxidation in M1 macrophages. M1 macrophages were transfected with siRNA negative control or si-fatty acid desaturase 1 for 48 hours, and lipid peroxidation was assessed using BODIPY™ 581/591 C11 staining. Representative fluorescence images show reduced green fluorescence (oxidized C11-BODIPY) and increased red fluorescence (reduced C11-BODIPY) in si-fatty acid desaturase 1-treated cells compared with those in siRNA negative controls, indicating decreased lipid peroxidation. FADS: Fatty acid desaturase; NC: Negative control.

FADS1 knockdown in M1 macrophages enhances proliferation inhibition of esophageal cancer cells

The results of Transwell coculture assays revealed that OE-19 esophageal carcinoma cells were very robust to proliferation under monoculture conditions. However, when OE-19 cells were cocultured with M1 macrophages, OE-19 cell growth was strongly suppressed (P < 0.0001). The growth inhibitory effect was stronger after we genetically silenced FADS1 in M1 macrophages. Quantitatively, the viability (as measured by the OD) of OE-19 cells cocultured with FADS1-depleted macrophages was much lower than that of the negative control group at the 48-, 72- and 96-hour time points (P < 0.001) (Figure 6A). These findings indicate that FADS1 suppression can maintain and enhance the tumor-suppressive function of M1 macrophages by modulating ferroptosis.

Figure 6 Fatty acid desaturase 1 promotes M1 macrophage ferroptosis via 1-palmitoyl-2-stearoyl-glycerophosphocholine (16:0/18:0). A: OE-19 cells were cultured alone or cocultured with M1 macrophages transfected with siRNA negative control or si-fatty acid desaturase 1-1 in a transwell system for 96 hours. Cell viability was assessed by a cell counting kit-8 assay at the indicated time points and expressed as optical density at 450 nm values. The data are shown as the mean ± SD of three independent experiments (n = 6 replicates per group); B and C: M1 macrophages were transfected with siRNA negative control or si-fatty acid desaturase 1-1 and treated with or without 10 μM phosphatidylcholine (16:0/18:0) for 24 hours, after which the malondialdehyde concentration was measured. The data are presented as the mean ± SD from three independent experiments. FADS: Fatty acid desaturase; MDA: Malondialdehyde; OD₄₅₀: Optical density at 450 nm; PC: Phosphatidylcholine.

FADS1 promotes ferroptosis by specifically regulating 1-palmitoyl-2-stearoyl-GPC (16:0/18:0)

To determine which lipid metabolite mediates FADS1-triggered ferroptosis, phospholipid restoration assays were also performed. As shown in the experiments, the administration of exogenous 1-palmitoyl-2-stearoyl-GPC (16:0/18:0) (phosphatidylcholine treatment) strongly increased the MDA concentration in M1-polarized macrophages (0.617 ± 0.009 nmol/mL, P < 0.0001). When this phospholipid was added to FADS1-deficient cells (si-FADS1 + phosphatidylcholine condition), the reduction in MDA production was reversed by the suppression of FADS1, and the level of MDA returned to 0.482 ± 0.018 nmol/mL, which was obviously greater than that in the si-FADS1 alone group (P < 0.0001) (Figure 6B and C). These results suggest that FADS1 promotes lipid peroxidation and thus ferroptosis at least in part by increasing the 1-palmitoyl-2-stearoyl-GPC (16:0/18:0) level.

FADS1 promotes lipid raft formation via the above phospholipid and induces ferroptosis in a lipid raft-dependent manner

Immunofluorescence analysis revealed that the suppression of FADS1 expression resulted in a decrease in the signal strength of caveolin-1, which is a signal of the presence of lipid rafts, indicating the impairment of raft assembly. On the other hand, the administration of 1-palmitoyl-2-stearoyl-GPC (16:0/18:0) significantly increased the level of caveolin-1 expression (Figure 7). Further experiments with the lipid raft disruptor methyl-β-cyclodextrin treatment revealed that this compound reversed the phospholipid-induced increase in lipid peroxidation, as demonstrated by the diminished green fluorescence signals and the recovered red fluorescence pattern (Figure 8). Therefore, the FADS1-phospholipid pathway initiates ferroptosis in M1 macrophages by facilitating the organization of lipid rafts.

Figure 7 Fatty acid desaturase 1 promotes lipid raft formation in M1 macrophages via 1-palmitoyl-2-stearoyl-glycerophosphocholine (16:0/18:0). M1 macrophages were divided into four groups: Control (siRNA negative control), fatty acid desaturase 1 knockdown (si-FADS1), phosphatidylcholine (PC) (16:0/18:0) treatment, and si-FADS1 combined with PC treatment (si-FADS1 + PC). Lipid raft formation was assessed by immunofluorescence staining for caveolin-1 (green), with nuclei counterstained with 4’,6-diamidino-2-phenylindole (blue). Representative fluorescence images show that caveolin-1 fluorescence intensity was significantly lower in the si-FADS1 group than in the siRNA negative control group, markedly increased in the PC treatment group, and partially restored by PC treatment in si-FADS1 cells (si-FADS1 + PC). FADS: Fatty acid desaturase; PC: Phosphatidylcholine; DAPI: 4’,6-diamidino-2-phenylindole.

Figure 8 1-palmitoyl-2-stearoyl-glycerophosphocholine promotes ferroptosis in M1 macrophages via lipid rafts. M1 macrophages were divided into four groups: Control (negative control), phosphatidylcholine (PC) (16:0/18:0) treatment, lipid raft inhibitor methyl-β-cyclodextrin treatment (MβCD), and combined treatment with PC and MβCD (PC + MβCD). Lipid peroxidation was assessed using the BODIPY™ 581/591 C11 probe, with the nuclei counterstained with Hoechst 33342. Representative fluorescence images show that control cells exhibited moderate green fluorescence (oxidized C11-BODIPY, O-C11), indicating basal lipid peroxidation. PC treatment markedly increased green fluorescence and decreased red fluorescence (reduced C11-BODIPY, R-C11), reflecting elevated lipid peroxidation. MβCD treatment alone reduced green fluorescence and enhanced red fluorescence, suggesting the inhibition of lipid peroxidation. Notably, compared with the PC treatment, the MβCD treatment significantly reversed the PC-induced increase in lipid peroxidation, as evidenced by decreased O-C11 BODIPY and increased R-C11 BODIPY fluorescence. PC: Phosphatidylcholine.

FADS1 virtual knockout analysis reveals a potential target gene network regulating myeloid cell functions

Computational analysis using the virtual knockout approach of scTenifoldKnk revealed large changes in the gene regulatory networks in monocytes/macrophages after the depletion of FADS1, and most of the genes whose expression changed were positively regulated (Z score greater than zero), as shown in Figure 9A. The genes whose expression changed the most were MZB1, SPAG4, ITGB2, LILRB4, and LAIR1, and these were among the top 20 differentially regulated targets (Figure 9B). Of particular interest are the ITGB2, LILRB4 and SIRPA genes, which are known to be involved in immune suppression mechanisms, phagocytic activity modulation and myeloid lineage development. This same regulatory network also includes important myeloid-specific transcription factors and effector molecules such as SPI1, CTSS and TYROBP. These findings indicate that FADS1 potentially plays a wide role in both the development of myeloid cells and the modulation of the immune system through this gene network.

Figure 9 Impact of virtual fatty acid desaturase 1 knockout on the gene regulatory network in monocytes/macrophages. A: Scatter plot showing the distribution of differentially expressed genes following fatty acid desaturase 1 knockout. The X-axis represents the Z score (differential regulation score), and the Y-axis represents the negative logarithm of the adjusted P value. Positive Z scores indicate upregulated regulatory activity after knockout (positive regulation), whereas negative values indicate downregulation. The dashed line indicates the significance threshold (adjusted P value < 0.05); B: Bar plot displaying the top 20 genes most significantly affected by fatty acid desaturase 1 knockout along with their corresponding Z scores. FADS: Fatty acid desaturase.

Artificial intelligence-assisted drug screening and molecular docking identify potential FADS1-targeting small molecules

We used the DrugReflector computational model to study the genome-wide v-score profile for drug screening and identified several interesting compounds that can be used to reverse the transcriptional changes that we observed in esophageal cancer (Figure 10A). The BRD-K95779464 compound had the greatest transcriptomic reversal potential because its logit score was greater than 0.01. Computational modeling of the molecular interaction revealed that the compound is stably inserted in the hydrophobic binding cavity of FADS1, with a good Vina docking score of -8.3 kcal/mol, which is well below the -5.0 kcal/mol threshold for good binding (Table 3). On the basis of the detailed structural analysis, we detected both the formation of hydrophobic contacts and the hydrogen bonding pattern between the compound and important residues of FADS1, such as Phe48 and Ser188, in the catalytic site of FADS1 (Figure 10B). Taken together, these findings identify BRD-K95779464 as a possible FADS1-targeting therapeutic candidate, with new possibilities for an intervention strategy for esophageal cancer.

Figure 10  Artificial intelligence-assisted drug screening and molecular docking analysis of fatty acid desaturase 1-targeting compounds. A: Distribution of logit scores and probabilities for drugs predicted to reverse the esophageal cancer transcriptional signature by the DrugReflector model, highlighting the top 10 candidate drugs with the highest reversal scores; B: Molecular docking analysis showing the binding mode of BRD-K95779464 with the fatty acid desaturase 1 protein (left: Overall binding conformation; right: Close-up view displaying key contact residues).

Table 3 Molecular docking results of the candidate compound BRD-K95779464 with the fatty acid desaturase 1protein.

CurPocket ID	Vina score	Cavity volume (ų)	Center (X, Y, Z)	Docking size (X, Y, Z)
C1	-8.3	5051	(-4, 8, -1)	(28, 28, 35)
C5	-8.1	163	(-8, -12, 13)	(22, 22, 22)
C2	-7.6	358	(-2, -5, -16)	(22, 22, 22)
C3	-6.2	203	(14, 13, 9)	(22, 22, 22)
C4	-5.7	186	(7, 8, -23)	(22, 22, 22)

DISCUSSION

Esophageal carcinoma is an extremely invasive digestive tract malignancy with a poor clinical prognosis. The etiology of the disease involves complex interactions among genetic predisposition, environmental influences and metabolic dysregulation[1,2]. Currently, immune cell dynamics in the tumor microenvironment are also very much involved in the progression of esophageal cancer. Among these immune components, macrophages have dual functions: They can suppress tumor growth or can favor tumor progression, producing cytokines and modulating the immune response[3]. Moreover, disturbances in lipid metabolic pathways are well described in several types of cancer, and these alterations are strongly related to the proliferation of malignant cells, invasion capacity and immune system avoidance mechanisms[18,19]. The enzyme FADS1, which is a FADS, strongly regulates the synthesis of polyunsaturated fatty acids and is related to the metabolic reprogramming of tumor cells[5,20]. Genetic polymorphisms in this enzyme are strongly correlated with susceptibility to digestive tract malignancies. Therefore, performing detailed research on the role of FADS1 in esophageal carcinoma and its tumor-surrounding microenvironment, particularly the pathways related to ferroptosis in immune cells, is highly important for understanding the pathogenesis and progression of this condition.

Notably, a bidirectional MR analysis revealed that the OR for esophageal cancer associated with FADS1-related genetic variants at the genetic level was less than 1, suggesting that an increased genetic background of FADS1 serves as a protective factor against esophageal cancer susceptibility. These findings contrast with results from clinical tissue samples and transcriptomic data showing significant upregulation of FADS1 in tumor lesions; however, the two approaches differ in their perspectives. MR is based on population-level genomic variants and reflects the genetic relationship between FADS1 and esophageal epithelial carcinogenesis under systemic homeostatic conditions. In contrast, quantitative transcriptomic and immunofluorescence data analyses revealed significantly higher FADS1 mRNA and protein expression in tumor tissues than in paired adjacent nontumor tissues. Single-cell sequencing further revealed that FADS1 upregulation is primarily enriched in tumor-infiltrating M1-type macrophages, reflecting pathological local overexpression resulting from tumor microenvironment remodeling. Following tumorigenesis, inflammation, hypoxia, and cytokine dysregulation within the lesion collectively drive abnormal upregulation of FADS1 expression in M1 macrophages. Excess FADS1 leads to the abundant synthesis of 1-palmitoyl-2-stearoyl-GPC (16:0/18:0), which remodels the phospholipid profile of macrophages, promotes lipid raft assembly in cell membranes, and induces ferroptosis in M1 macrophages. This directly impairs the anti-tumor function of M1 macrophages in eliminating esophageal cancer cells, ultimately exerting a pro-carcinogenic effect within the local tumor microenvironment.

In this research, we performed a comprehensive study on the presence and expression of FADS1 in esophageal carcinoma specimens and tumor-infiltrating macrophages and investigated how FADS1 affects the lipid metabolism pathway in M1-polarized macrophages. Single-cell transcriptome analysis, immunofluorescence imaging and functional assays revealed that FADS1 can promote lipid peroxide accumulation and increase ferroptosis-related cell death in M1 macrophages, thereby impairing their antitumor activity and promoting the growth of malignant esophageal cells. These results provide new insights into FADS1, which is an important mediator between lipid metabolism and the immune response in the tumor microenvironment. We next examined how FADS1 remodels lipid metabolic networks, alters membrane microdomain organization, and affects macrophage function.

FADS1 was shown to be overexpressed in esophageal carcinoma specimens, predominantly localized in tumor-infiltrating macrophages. Genetic silencing of FADS1 led to a marked reduction in the level of lipid peroxide in M1-polarized macrophages, indicating its important role in the process of ferroptosis cell death through the modulation of lipid peroxidation. These findings are consistent with those of previous studies showing that the enzymatic function of FADS1 is to act as an acyl-CoA desaturase that regulates the biosynthesis of polyunsaturated fatty acids, therefore affecting cellular membrane lipid profiles and oxidative stress pathways[5]. Notably, lipid peroxidation is a fundamental mechanism of ferroptosis. The enzymatic activity of FADS1 can increase the proportion of oxidation-susceptible polyunsaturated fatty acids among membrane phospholipids, thereby increasing lipid peroxidation and the subsequent events of ferroptosis[21]. In previous works, only the metabolic regulatory functions of FADS1 in malignant cells were studied, and this work is the first to demonstrate the ferroptosis-promoting activity of FADS1 in tumor-associated macrophages. This study showed how FADS1 regulates lipid metabolism in tumor-infiltrating macrophages and revealed a novel route through which this enzyme changes the behavior of immune cells. This study contributes to our understanding of the different regulatory functions of FADS1 in the tumor microenvironment and highlights the importance of this enzyme in controlling the activity of macrophages and ferroptosis processes. This study provides new evidence of the complex mechanisms of metabolic reprogramming that shape the immune response in cancerous tissue.

Further investigations indicated that FADS1 is involved in the formation of lipid rafts since it participates in the production of particular phospholipid species, e.g., 1-palmitoyl-2-stearoyl-glycerophosphocholine (16:0/18:0). This biochemical process then regulates ferroptosis-related pathways. These specialized membrane microdomains, which contain high concentrations of cholesterol and saturated fatty acids, are organization centers for the clustering of signaling molecules. The fluidity of these structures strongly influences the functionality of immune cells[22]. By pharmacological disruption of lipid rafts and measuring the level of lipid oxidation, this work demonstrated the important function of these lipid rafts in the control of FADS1-mediated ferroptosis. These results suggest that the role of FADS1 involves not only simple modulation of the lipid profile but also structural reorganization of the cell membrane and therefore the influence of signaling cascades in macrophages. Although some studies have investigated how lipid metabolism can affect the activity of immune cells, the present work provides more precise information. This study revealed how FADS1 regulates lipid-derived metabolites to regulate the architecture of the membrane and thus plays an important role in the processes of ferroptosis. It also addresses how FADS1 affects membrane signaling microenvironments. It further provides basic molecular information that helps elucidate the complex relationship between lipid metabolism and immune system regulation.

In terms of cellular interactions, M1 macrophages with reduced FADS1 expression cocultured with esophageal carcinoma cells exhibited marked suppression of malignant cell growth. These results suggest that FADS1 plays an important role in enhancing macrophage-mediated tumor suppression through the modulation of ferroptosis. The observed effects are in accordance with the scientific literature that has documented the correlation of macrophage polarization states with their ability to control neoplastic cell expansion. Indeed, as a different mode of regulated cell death, ferroptosis has been shown to affect the immune effector capabilities of macrophages[21,23]. FADS1 also seems to modulate the concentrations of reactive oxygen species and proinflammatory cytokine secretion in macrophages through lipid metabolic pathways, thus affecting immune regulation in the tumor niche[24]. The results of the present investigation revealed the exact molecular pathways through which FADS1 orchestrates lipid metabolism alterations and ferroptosis signaling that govern macrophage-mediated tumor suppression. This study was the first to reveal the crucial role of FADS1 in the antitumor function of macrophages at the molecular level, advancing our understanding of the regulation of immune metabolism in tumor progression.

Transcriptomic profiling at single-cell resolution revealed differential enrichment of FADS1 in macrophages in esophageal carcinoma, with the expression levels strongly correlated with macrophage functional phenotypes. These findings suggest that FADS1 potentially plays a modulatory role in the dynamics of the immune landscape of tumors. Previous studies also support the involvement of FADS1 in lipid metabolic pathways and in the regulation of immune cells in different tumor types and in different tissue contexts[24,25]. Our study extends beyond these observations by revealing the heterogeneous expression patterns of FADS1 in tumor-infiltrating macrophages and its connection to cellular functionality, which suggests the potential role of FADS1 in the influence of immune cell polarization and activation through the regulation of lipid metabolism. In previous studies, the functions of FADS1 were mainly related to malignant cells, whereas in the present study, the expression profile and biological significance of FADS1 were more strongly related to immune populations, providing new perspectives on the lipid-mediated regulatory mechanism in the tumor-immune microenvironment.

Biochemical analysis with the use of the BODIPY™ 581/591 C11 fluorescent indicator revealed a strong reduction in the lipid peroxidation of M1 macrophages after the silencing of the FADS1 gene. These findings indicate that FADS1 plays important roles in the regulation of lipid homeostasis and metabolic pathways associated with ferroptosis. The process of lipid peroxidation, which is a hallmark of ferroptosis, is closely linked to the metabolism of polyunsaturated fatty acids. FADS1 is an enzyme involved in the desaturation of fatty acids and plays an important role in the determination of the oxidative vulnerability of the lipids of the cellular membrane[10,26]. Moreover, the variation in the level of lipid peroxidation has direct consequences for cellular redox equilibrium and for the functionality of programmed cell death pathways[21]. By using a special probe to monitor the variation in the level of lipid peroxidation in real time, this work provides direct biochemical evidence for the FADS1 participation in the regulation of ferroptosis through the regulation of lipid metabolism. This finding is more convincing than previous indirect inferences based on gene expression and metabolomics and highlights the central role of FADS1 in regulating lipid oxidation homeostasis in macrophages.

The results of this study involving whole-genome metabolomics analysis revealed the important role of FADS1 in maintaining the lipid oxidation balance of macrophages. Nevertheless, some limitations of this investigation should be considered. The main limitations of the study are the small number of participants and the lack of human clinical specimens, which may hamper the generality of the results obtained. Even with very advanced methodologies, such as single-cell transcriptome profiling, most of the experimental data obtained are from in vitro cellular models and do not properly represent the complex tumor ecosystem. Moreover, in this study, only THP-1-derived macrophages and a single OE-19 esophageal cancer cell line were used for in vitro functional validation, and verification in primary macrophages and multiple esophageal cancer cell lines was lacking. Differences between experimental batches may reduce data consistency across different tests. Future work will employ patient-derived primary macrophages and additional esophageal cancer cell lines to further confirm the universality of the proposed regulatory axis. In addition, the identification of the candidate inhibitor BRD-K95779464 relies entirely on computational artificial intelligence screening and molecular docking simulations, without any experimental validation. Therefore, this compound should be considered only a preliminary computational screening result rather than a confirmed FADS1 inhibitor.

This investigation is the first to reveal that FADS1 is important for the progression of esophageal carcinoma; more specifically, FADS1 functions to control the immune reaction of tumors by regulating ferroptosis in M1 macrophages. A more complete study of how FADS1 works could open new approaches in oncology, particularly in the cellular environment of tumors, where FADS1 could be an important controlling element. Further studies should focus on the manipulation of FADS1 and related molecular mechanisms to develop better methods for targeted esophageal cancer therapy.

CONCLUSION

FADS1 assembles lipid rafts by reprogramming the phospholipid metabolism, thereby triggering ferroptosis in M1 macrophages and promoting the progression of esophageal cancer. Specifically, FADS1 is highly expressed in esophageal cancer tissues and tumor-infiltrating M1-type macrophages. It promotes the assembly of lipid rafts through the accumulation of specific phospholipid molecules, such as 1-palmitoyl-2-stearoyl-GPC (16:0/18:0), inducing ferroptosis in M1 macrophages and thereby weakening their antitumor ability. Inhibiting FADS1 expression can reduce lipid peroxidation, suppress ferroptosis, and increase the ability of M1 macrophages to kill tumor cells. Additionally, computer simulation revealed that the compound BRD-K95779464 is a promising FADS1 inhibitor.

ACKNOWLEDGEMENTS

The authors thank all the participants and staff involved in this study for their contributions.