FADS1-mediated Efferocytosis in Tumor-Associated Macrophages Shapes an Immunosuppressive Microenvironment in Gastric Cancer

Abstract

Tumor-associated macrophages (TAMs) play a central role in tumor progression and therapeutic resistance. Alterations in unsaturated fatty acids (UFAs) metabolism are known to contribute to the immunosuppressive properties of TAMs. However, the complex regulatory relationship between UFAs metabolism and TAMs during gastric cancer (GC) progression remains largely unclear. Through a comprehensive analysis of single-cell RNA sequencing and transcriptomic data, we found that the UFAs biosynthesis-related enzyme FADS1 holds prognostic significance in GC within TAMs. Clinically, a high infiltration of FADS1⁺TAMs strongly correlates with poor prognosis and immunotherapy resistance across several GC cohorts. Using in vitro co-culture systems and in vivo mouse models, our data suggest that FADS1⁺TAMs actively contribute to an immunosuppressive microenvironment, potentially driving CD8⁺T cell dysfunction. Mechanistically, our findings indicate that FADS1 mediated DHA synthesis and activated PPARγ, which subsequently promotes efferocytosis and inhibit the STING-IFN-I signaling pathway in TAMs. Notably, systemic pharmacological targeting of FADS1 exerted a significant sensitizing effect in mouse ectopic GC models, inhibiting tumor growth, enhancing CD8⁺T cells cytotoxicity, and improving the efficacy of PD-1 blockade. Overall, our study highlights FADS1⁺TAMs as a critical component of GC progression and a potential exploratory biomarker for immunotherapy responsiveness. Furthermore, pharmacological inhibition of FADS1 represents a promising combinatorial strategy for reshaping the TME and enhancing immune checkpoint blockade efficacy in preclinical settings.

Supplementary Information

The online version contains supplementary material available at https://doi.org/10.1007/s00018-026-06314-x.

Introduction

Gastric cancer (GC) is one of the most prevalent and aggressive malignancies worldwide, posing a significant threat to human health. According to the latest global cancer statistics released by the International Agency for Research on Cancer in 2022, GC ranks as the fifth most common cancer and the third leading cause of cancer-related deaths globally [1]. D2 radical surgery combined with perioperative comprehensive treatment has improved the prognosis of patients with locally advanced gastric cancer (LAGC) [2]. However, the overall response to treatment remains limited, with recurrence and metastasis rates still relatively high, hindering further improvements in long-term survival [3]. Immune checkpoint inhibitors (ICIs)-based immunotherapy has emerged as a new benchmark for the first-line treatment of advanced GC. Nevertheless, a significant portion of LAGC patients exhibit resistance to immunotherapy. A deeper understanding of the immunosuppressive characteristics of the GC tumor microenvironment (TME) is crucial for identifying novel and more effective immunotherapeutic strategies [4].

Tumor-associated macrophages (TAMs) are the predominant innate immune cells within TME and play a key role in immune therapy resistance and immune evasion, particularly by inhibiting T-cell recruitment and immune responses [5]. Understanding how TAMs acquire tumor-promoting functions is essential for overcoming T-cell immunosuppression. Advances in single-cell RNA-sequencing technology have revealed significant heterogeneity among TAMs subpopulations in tumor tissues [6]. The remarkable plasticity of TAMs, defined by their diverse phenotypic, metabolic, and functional profiles, highlights their role in immune suppression while also offering potential opportunities for therapeutic intervention [7]. Deciphering the molecular signatures of TAMs and identifying key subsets are critical for reversing the immunosuppressive state of the TME and enhancing the efficacy of immunotherapy.

The biosynthesis of unsaturated fatty acids (UFAs) plays a critical role in tumor proliferation, metastasis, and chemoresistance [8]. Emerging evidence suggests that disruptions in UFAs metabolism, in concert with the TME, profoundly influence key cancer hallmarks. These alterations drive tumor progression and therapeutic resistance by promoting inflammation that supports tumor growth, enhancing angiogenesis, modulating stromal cell behavior, and facilitating immune evasion through significant changes in the immune cell landscape, especially in TAMs [9]. Lin et al. reported that tumor-derived UFAs enhance the immunosuppressive properties of TAMs through the FABP5-PPARγ signaling pathway. Targeted inhibition of FABP5 has been shown to improve immune checkpoint blockade therapy [10]. In addition, Chan et al. found that abundant UFAs promote protumoral properties in TAMs by downregulating the RhoA-GTPase-YAP1 signaling pathway within the ovarian cancer TME [11]. However, due to the complexity of the extensive UFA regulatory system, it remains unclear whether UFA metabolic enzymes and their products in TAMs play a role in regulating GC progression.

Fatty acid desaturase 1 (FADS1) is a pivotal enzyme in UFAs metabolism, responsible for introducing double bonds into fatty acid molecules. This process converts saturated fatty acids into unsaturated fatty acids, playing a critical role in the synthesis of long-chain polyunsaturated fatty acids (PUFAs) in the ω-6 and ω-3 series [12]. Research has shown that FADS1 regulates UFAs metabolism and modulates ferroptosis susceptibility in GC and triple-negative breast cancer (TNBC) [13, 14]. The FADS1-arachidonic acid axis enhances UFAs metabolism by modulating intestinal microecology in colorectal cancer [15]. Notably, the orally available, potent, and selective inhibitor of FADS1, D5D-IN-326 (Compound-326), has been shown to reduce insulin resistance, body weight, and atherosclerosis progression in animal models, highlighting its potential as a therapeutic target for cancer [16]. However, previous research has mainly concentrated on FADS1 within tumor cells, with its role in the TME still largely underexplored. Therefore, a thorough investigation of how FADS1 impacts the TME is essential for translating these findings into clinical applications. This could potentially improve therapeutic strategies and offer a deeper insight into tumor biology, particularly in relation to immune regulation.

In this study, we identified a previously unrecognized association between FADS1 and GC progression. Our data suggest that FADS1-positive TAMs are closely involved in the establishment of an immunosuppressive TME. Furthermore, our clinical evaluation revealed a significant correlation between FADS1⁺TAM infiltration, therapeutic response, and clinical outcomes in GC patients. Mechanistically, our findings indicated that FADS1 may enhance efferocytosis by activating PPARγ through DHA synthesis in TAMs, thereby inhibiting the STING-IFN-I pathway and impairing CD8⁺T cell-mediated anti-tumor immunity. Overall, this study highlights the potential role of FADS1 in promoting an immunosuppressive phenotype, offering new insights into the molecular mechanisms of immune evasion and suggesting novel strategies for targeting TAMs to enhance T-cell-mediated antitumor responses.

Materials and methods

Clinical cohorts

Tumor specimens from PUMCH cohort I (n = 80), PUMCH cohort II (n = 70), PUMCH cohort III (n = 40), and PUMCH IV (n = 13), consisting of GC patients who underwent gastrectomy between January 2015 and December 2017 at the Department of General Surgery, Peking Union Medical College Hospital, were used for analysis. Patients were followed up at 6-month intervals until December 2022. Overall survival (OS) was calculated from the date of surgery to the date of death or the last follow-up. The histopathological tumor response was evaluated using a simplified four-tiered tumor regression grading (TRG) system recommended by the College of American Pathologists (CAP-TRG) [17]. Treatment response to NT was categorized into two groups: non-responders (CAP 2–3) and responders (CAP 0–1). Informed consent was obtained from all patients, and the study protocol was approved by the Institutional Review Board of Peking Union Medical College Hospital.

Publicly transcriptomic data collection and processing

The RNA-seq data from the TCGA-STAD (Stomach Adenocarcinoma) project were downloaded, organized using the STAR pipeline from the TCGA database (https://portal.gdc.cancer.gov), and the data were extracted in TPM format. Expression data from paired GC and normal tissues were obtained from GSE63089 on the Gene Expression Omnibus (GEO). Differential expression of FADS1 between tumor and normal tissues was analyzed and visualized using the “ggpubr” package in R. The prognostic analysis of UFAs biosynthesis-related genes in the TCGA-STAD datasets was performed using the GEPIA 3 database.

The TCGA cohort included 370 GC patients with available survival data, with both gene expression profiles and corresponding clinical information retrieved from the UCSC XENA website (https://xenabrowser.net/datapages/). Additionally, gene expression profiles and clinical data from the GSE26942 cohort (n = 96, non-adjuvant therapy), sourced from the GEO database, were incorporated into this study for further analysis. To evaluate overall macrophage abundance, the CIBERSORT algorithm was employed to quantify the relative fractions of macrophage subtypes (M0, M1, and M2) within the bulk RNA-seq cohorts. Furthermore, the Tumor Immune Dysfunction and Exclusion (TIDE) score was utilized to evaluate the potential clinical efficacy of immunotherapy across different risk groups, serving as an indicator of the underlying tumor immune evasion capacity (http://tide.dfci.harvard.edu).

Single-cell RNA sequencing data analysis

Single-cell RNA sequencing (scRNA-seq) data of GC patients were obtained from the GSE183904 dataset in the GEO database. A standard workflow from the “Seurat” package in R was employed to perform quality control, data normalization, batch effects correction, dimensional reduction, and cell clustering. Uniform Manifold Approximation and Projection (UMAP) reduction was then applied for cell visualization and analysis. Differentially expressed genes (DEGs) within the clusters were identified using the “FindAllMarkers” function. Cluster-specific cell lineages were annotated by cross-referencing marker genes with the CellMarker and PanglaoDB databases for more accurate identification.

To define the FADS1⁺TAMs signature, TAMs were first extracted from the scRNA-seq data and then categorized into FADS1⁺TAMs and FADS1⁻TAMs based on FADS1 expression levels. DEGs between these two groups were identified using the “FindMarkers” function. To estimate the infiltration levels of FADS1⁺TAMs in bulk RNA sequencing (bulk-RNA seq) data, the average expression of the top 20 unique markers derived from FADS1⁺TAMs was calculated, which was then used to generate the FADS1⁺TAMs signature score [18].

Differential expression (DE) analysis, Gene Ontology (GO) analysis, Kyoto Encyclopedia of Genes and Genomes (KEGG) analysis, and gene set enrichment analysis (GSEA) were performed using the “msigdbr”, “clusterProfiler”, “dplyr”, and “ggplot2” packages in R. To calculate the cell-level activity scores of gene sets, the AUCell function from the “Seurat” package was utilized.

Cell culture and transfection

The human monocyte cell line THP-1, human GC cell line MKN-45, and mouse GC cell line mouse forestomach carcinoma (MFC) were purchased from the National Infrastructure of Cell Line Resource (NICR, China). Cells were cultured in RPMI-1640 medium (Gibco) supplemented with 10% fetal bovine serum (FBS; Gibco) and 1% penicillin/streptomycin (Gibco) in a humidified atmosphere of 5% CO₂ at 37℃. In addition, bone marrow-derived macrophages (BMDMs) extracted from the thighbones of 615-line mice were cultured in the Dulbecco’s modified Eagle medium (DMEM; Gibco) for 7 days, supplemented with 10% FBS, 1% penicillin/streptomycin, and M-CSF (25 ng/ml, PeproTech), to prepare for further experiments.

Lentiviruses carrying plasmids overexpressing (OE-FADS1, Ubi-MCS-3FLAG-CBh-gcGFP-IRES-puromycin) or knocking out FADS1 (KO-FADS1, U6-sgRNA-EF1a-Cas9-FLAG-CMV-EGFP-P2A-puro) were designed and constructed by GeneChem (Shanghai, China). Lentiviruses were infected into THP-1 cells following the manufacturer’s instructions. The cells were then selected with puromycin (2 µg/mL) and sorted using fluorescence-activated cell sorted (FACS) to establish stable cell lines for subsequent experiments. For knockdown assays, the FADS1 siRNA and PPARγ siRNA (hippobio, Huzhou, China) were transfected into the BMDMs or THP-1 using RNAiMAX Transfection Reagent (Lipofectamine).

Co-Culture Experiments

THP-1 cells were induced to differentiate into macrophages by stimulation with phorbol 12-myristate 13-acetate (100 ng/mL, Sigma-Aldrich) for 24 h, after which they were used in subsequent experiments. To obtain TAMs, MKN-45 cells were seeded into the upper chamber, while macrophages (FADS1 overexpressed or knocked out) were cultured in the bottom chamber with a 0.4-µm insert (Corning). The cells were co-incubated for 48 h.

Fresh human peripheral blood samples were collected in anticoagulation tubes, and peripheral blood mononuclear cells (PBMCs) were isolated by density gradient centrifugation using lymphocyte separation medium (Solarbio). T cells were isolated by flow cytometry and co-cultured with TAMs at a concentration of 2 × 10⁵ cells/mL for 48 h in RPMI-1640 medium supplemented with 10% FBS and IL-2 (10 ng/mL, PeproTech). The STING agonist diABZI (100 nM, MedChemExpress) was added to the co-culture system. After 6 h of treatment with Cell Stimulation Cocktail (plus protein transport inhibitors) (Invitrogen), all cells were collected for flow cytometry analysis.

Establishment of mouse tumor model and treatment

The 615-line mice, syngeneic hosts for the MFC cell line, were purchased from the Institute of Hematology, Chinese Academy of Medical Sciences (Tianjin, China). To establish a subcutaneous tumor model, 5 × 10⁵ MFC cells and 2.5 × 10⁵ BMDM cells were subcutaneously injected into the right armpit of 5-week-old male 615-line mice. Tumor volumes were calculated using the modified ellipsoidal formula: volume = length×width²/2. All mice were monitored every day.

For the drug treatment experiment, subcutaneous GC models were first established by injecting 5 × 10⁵ MFC cells into the right armpit of 5-week-old male 615-line mice. Three days after inoculation, the mice were randomly assigned to one of four treatment groups: intraperitoneal injection of anti-PD-1 (10 mg/kg every 2 days, Leinco Technologies), intragastric administration of the FADS1 inhibitor D5D-IN-326 (10 mg/kg daily, Sigma-Aldrich), a combination of anti-PD-1 and D5D-IN-326, or a control group. The mice were sacrificed 15 days after the treatment, and tumors were harvested and weighed. We strictly followed ethical principles for animal care and obtained approval from the Institutional Animal Care and Use Committee at the Shanghai Yishang Biotechnology Center.

Flow cytometry analysis

Single-cell suspensions were surface-stained on ice for 20 min in the dark with fluorochrome-conjugated antibodies, including FITC anti-human CD3 antibody and PE/Cyanine7 anti-human CD8 antibody. After fixation, the cells were resuspended in permeabilization wash buffer and intracellularly stained with PE anti-human IFN-γ antibody and APC anti-human/mouse Granzyme B recombinant antibody. All reagents were purchased from BioLegend. Flow cytometric analysis was performed using the CytoFLEX S Flow Cytometer (Beckman Coulter). Data were collected and analyzed using FlowJo software.

For animal experiments, fresh mouse tumors from 615-line mice were harvested and processed into single-cell suspensions through mechanical dissociation. The cells were digested with tissue dissociation buffer (Absin) and incubated at 37 °C with shaking at 180 rpm for 40 min. The cells were then filtered through 70-µm nylon cell strainers (BD Falcon) and washed twice with PBS. After centrifugation and removal of the supernatant, the cells were resuspended in PBS and incubated with the Zombie Red Fixable Viability Kit before surface labeling with antibodies. Subsequently, the cells were incubated with pre-conjugated surface antibodies (PerCP/Cyanine5.5 anti-mouse CD45 antibody, APC/Cyanine7 anti-mouse CD3 antibody, and PE/Cyanine7 anti-mouse CD8a antibody), followed by fixation, permeabilization, and intracellular antibody incubation with PE anti-mouse IFN-γ antibody and APC anti-human/mouse Granzyme B recombinant antibody. All reagents were purchased from BioLegend.

Western blotting assay, immunohistochemistry, and immunofluorescence

For Western blotting, total protein was lysed and extracted using RIPA Lysis and Extraction Buffer (Thermo Scientific), supplemented with a Protease and Phosphatase Inhibitor Cocktail (Thermo Scientific). Protein samples were separated by 4–12% SDS-PAGE (Sangon Biotech) and transferred onto PVDF membranes (Millipore). Membranes were blocked using 5% non-fat milk (Sangon Biotech) and incubated overnight at 4 °C with primary antibodies: FADS1 (Atlas, HPA042705), FADS1 (Proteintech, 10627-1-AP), STING (CST, #13647), TBK1 (CST, #3504), IRF3 (CST, # 4302), phosphor-STING (CST, #19781), phosphor-TBK1 (CST, #5483), phosphor-IRF3 (CST, #4947), and ACTB (CST, #4970). After rinsing, membranes were incubated with corresponding HRP-conjugated secondary antibodies (CST, #7076, #7074), and antigen-antibody reactions were visualized using the enhanced chemiluminescence assay (Thermo Scientific). All pathway-related primary antibodies, internal control antibody, and secondary antibodies were purchased from Cell Signaling Technology (CST).

For immunohistochemistry, tissue sections were dewaxed, hydrated, and subjected to antigen retrieval. After blocking with 3% BSA, sections were incubated overnight at 4℃ with primary antibodies against FADS1 (Atlas, HPA042705), cleaved caspase-3 (CST, #9661), IFN-γ (huabio, ET1703-17), and Granzyme B (CST, #46890). Following incubation, slides were treated with horseradish peroxidase (HRP)-conjugated secondary antibodies for 1 h at room temperature. Color development was performed using DAB (Servicebio), and nuclear staining was carried out with hematoxylin (Servicebio). The stained slides were analyzed using CaseViewer software (3DHISTECH).

For immunofluorescence, tissue sections were dewaxed, hydrated, and subjected to antigen retrieval. Cells were fixed with 4% paraformaldehyde (Sangon Biotech) for 30 min. After blocking with 3% BSA, sections were incubated with primary antibodies against FADS1 (Atlas, HPA042705) and CD68 (Abcam, ab213363). Corresponding secondary antibodies conjugated with Alexa Fluor dyes were applied, and unbound antibodies were washed off using citrate buffer. For nuclear staining, DAPI (Servicebio) was used and incubated in a dark environment. Following staining, the 3DHISTECH fluorescence imaging scanner was employed for scanning, and quantitative analysis of the staining was conducted using the HALO platform.

RNA sequencing analysis

Total RNA was extracted from OE-FADS1 TAMs and OE-NC TAMs using TRIzol Reagent (Invitrogen). Subsequent procedures, including quality assessment, library construction, sequencing, and data analysis, were carried out using the Illumina X Plus platform (Novogene, Beijing, China). Three biological replicates were included for each condition in both the experimental and control groups. Differential expression analysis was performed using the “edgeR” R package to identify DEGs. Functional enrichment analysis, including GO, KEGG, and GSEA, was conducted to identify significantly enriched terms and pathways among the DEGs, with Bonferroni-corrected P-values < 0.05.

Real-time quantitative PCR (qPCR)

Total RNA was extracted from cells using the RNAeasy Animal RNA Isolation Kit with Spin Column (Beyotime), and reverse transcribed into cDNA using the PrimeScript RT reagent Kit (TaKaRa). Real-time qPCR was performed with the Applied Biosystems 7300 Plus Real-Time PCR System and TB Green Premix Ex TaqII (Tli RNaseH Plus) (TaKaRa). The relative expression of target genes was calculated using the 2^−ΔΔCt method, with ACTB as the internal control. Data were presented as fold change, normalized to the control group values. All oligonucleotide primers were synthesized by Sangon Biotech (Shanghai, China). The primer pairs for IFN-β were as follows: 5’-GCGACACTGTTCGTGTTGTC-3’(Forward) and 5’-GCCTCCCATTCAATTGCCAC-3’(Reverse). The primer pairs for OASL were as follows: 5’-CCATTGTGCCTGCCTACAGAG-3’(Forward) and 5’-CTTCAGCTTAGTTGGCCGATG-3’(Reverse). The primer pairs for IFIT1 were as follows: 5’-TTGATGACGATGAAATGCCTGA-3’(Forward) and 5’-CAGGTCACCAGACTCCTCAC-3’(Reverse). The primer pairs for Oas1 were as follows: 5’-TGTCCAAGGTGGTAAAGGGTG-3’(Forward) and 5’-CCGGCGATTTAACTGATCCTG-3’(Reverse). The primer pairs for ACTB were as follows: 5’-CCTGGCACCCAGCACAAT-3’(Forward) and 5’-GGGCCGGACTCGTCATAC-3’(Reverse).

Efferocytosis Assay

MKN-45 cells were treated with 100 µM 5-Fluorouracil (5-FU, Tianjin Jinyao) for 24 h (to prepare apoptotic cells), and then the cells were incubated with pHrodo iFL Red STP ester (Invitrogen) at room temperature for 1 h, protected from light. The stained apoptotic MKN-45 cells were added to TAMs, incubated for 2 h at 37 °C with 5% CO₂ [19]. Docosahexaenoic acid (DHA, 100µM, MedChemExpress) was added to the co-culture system for further experiments. The efferocytosis capacity of TAMs was analyzed by FACS analysis (Beckman Coulter).

Metabolomic analysis

Metabolites were isolated from the samples with a gradient of methanol solution. To normalize the data, quality control samples were made by combining equal parts from each experimental sample. The UHPLC-MS/MS system (Novogene) was used to separate and detect metabolites, employing both positive and negative ionization modes for thorough metabolite detection. To maintain quality assurance, QC samples were consistently injected throughout data collection. The raw data files from UHPLC-MS/MS were analyzed using XCMS software for peak alignment, peak selection, and metabolite quantification. Metabolites were identified as discriminatory using a Mann-Whitney U test, with those having a P-value below 0.05 deemed statistically significant.

Statistical analysis

All statistical analyses were conducted using GraphPad Prism 8.0.1 and R software 4.4.0. Quantitative data are presented as mean ± standard deviation (SD) or standard error of the mean (SEM) from at least three independent biological replicates. Comparisons between two groups were performed using the paired Student’s t-test or the Mann-Whitney test, as appropriate. For high-throughput transcriptomic analyses, P-values were adjusted for multiple comparisons using the Benjamini-Hochberg procedure to control the false discovery rate (FDR). Survival analysis was performed using the Kaplan–Meier method and evaluated with the log-rank test. Effect sizes are reported as Hazard Ratios (HR) with 95% confidence intervals (CI) for survival analyses. Detailed methodologies for specific assays are described in their respective sections above.

Results

FADS1 is highly expressed in TAMs and correlated with poor prognosis in GC patients

ScRNA-seq data revealed that the signature score of the UFAs biosynthesis pathway in TAMs was significantly higher in tumor tissues compared to normal tissues (Fig. 1A-C). To explore the key genes involved in UFAs metabolism among TAMs, we further analyzed DEGs between macrophages in tumor and normal tissues. This analysis identified five genes related to UFAs biosynthesis. Notably, analysis of the TCGA-STAD database revealed that only FADS1 had prognostic significance (Fig. 1D, E). IHC staining was performed on the PUMCH cohort I, and the results revealed that GC patients with higher FADS1 expression levels had a significantly shorter OS (Fig. 1F). The PUMCH cohort II was utilized for multiplex immunofluorescence staining to examine the expression of FADS1 and CD68 in GC tumor tissues. The results demonstrated the co-localization of FADS1 with CD68, indicating that FADS1 was expressed in TAMs. Furthermore, it was observed that high infiltration of FADS1⁺TAMs in GC patients was significantly associated with shorter OS and PFS (Fig. 1G).

Fig. 1.

Identification of FADS1⁺TAMs and their prognostic significance in GC.(A) UMAP plot represented single cells of GSE183904 dataset categorized into tissue types with distinct color codes. (B) UMAP plot represented single cells from tumor tissue in the GSE183904 dataset categorized into cell types with distinct color codes. (C) Violin plots showing the signature score of the UFAs biosynthesis pathway in macrophages from tumor tissues and normal tissues. (D) The left panel displays Venn diagrams illustrating the overlap of DEGs in macrophages and genes related to UFAs biosynthesis. The right panel highlights the intersection, identifying five genes and their prognostic value in TCGA-STAD. (E) Violin plots showing the expression of FADS1 in macrophages from tumor tissues and normal tissues. (F) Representative immunohistochemistry images of FADS1 expression in GC tissues (left), with the corresponding Kaplan–Meier analysis (right) from the PUMCH cohort I. Scale bars: 50 μm. (G) Representative multiplex immunofluorescence staining with anti-FADS1 and anti-CD68 antibodies, showing the co-localization of FADS1-expressing cells and TAMs in GC tissue. Scale bar: 100 μm (left). Kaplan–Meier survival curves for overall survival and progress-free survival in low and high FADS1⁺TAMs groups stratified in the PUMCH cohort II (right). (H) Kaplan–Meier survival curves for overall survival and progress-free survival in low and high FADS1⁺TAMs groups stratified in the PUMCH cohort III. (I, J) Survival analysis of GC patients with varying levels of FADS1⁺TAMs infiltration, based on the FADS1⁺TAMs signature in TCGA cohort (I) and GSE26942 cohort (J)

To verify the above results, we further validated the prognostic role of FADS1⁺TAMs in another independent cohort (PUMCH cohort III) (Fig. 1H). Further, we developed a scoring signature based on the mean expression of the top 20 markers derived from scRNA-seq data to assess the infiltration levels of FADS1⁺TAMs in bulk RNA-seq data. Survival analyses revealed that tumors with higher FADS1⁺TAMs signature scores were associated with significantly poorer OS in the TCGA and GSE26942 cohorts (Fig. 1I, J). To ensure these associations reflected the FADS1-specific TAMs state rather than a surrogate effect of overall macrophage abundance, we quantified general macrophage infiltration using the CIBERSORT algorithm. Crucially, multivariate Cox regression analysis demonstrated that the FADS1⁺TAMs signature remained a robust and independent prognostic factor for OS, even after adjusting for total macrophage abundance as a covariate (Supplementary Table S1, 2). Collectively, these findings suggest that FADS1⁺TAMs hold potential prognostic value for patients with GC.

Immunosuppressive impact of FADS1⁺TAMs on CD8⁺T cells in GC

To explore the role of FADS1⁺TAMs in GC, we next divided TAMs into FADS1⁺ and FADS1⁻ groups based on the expression of FADS1 from the scRNA-seq dataset. GSEA was subsequently performed on the DEGs between FADS1⁺TAMs and FADS1⁻TAMs to identify significantly enriched biological pathways. Our analysis revealed that pathways associated with fatty acid metabolism were enriched in the FADS1⁺TAMs, whereas immune activation pathways, including T cell mediated immunity and T cell mediated cytotoxicity, were suppressed in the FADS1⁻TAMs (Fig. 2A). We further constructed stable FADS1-overexpressing THP-1 cells, induced them into macrophages, and co-cultured them with GC cells to obtain TAMs in vitro (Fig. 2B). RNA-seq was performed on overexpressed-FADS1 (OE-FADS1) TAMs and the control. GSEA analysis showed a significant reduction in T cell-mediated immunity and T cell-mediated cytotoxicity pathways compared to the control (Fig. 2C).

Fig. 2.

Immunosuppressive impact of FADS1⁺TAMs on CD8⁺T cells in GC (A) Enrichment of immune and metabolism-related pathways in FADS1⁺TAMs. (B) Western blotting analysis of FADS1 in overexpressed-FADS1 (OE-FADS1) cells and negative control (OE-NC) derived from TAMs obtained through co-culture after differentiation of THP-1 cells. (C) GSEA plots of selected T cell-related gene sets in OE-FADS1 TAMs compared to OE-NC TAMs. (D) Flow chart of co-culturing TAMs and T cells to assess phenotypic changes in T cells (By Figdraw). (E, F) Expression profiles of IFN-γ (E) and GZMB (F) in CD8⁺T cells, quantified by flow cytometry in OE-FADS1 TAMs and KO-FADS1 TAMs. (G) Western blotting analysis of FADS1 expression in BMDMs derived from 615-line mice following siRNA transfection. (H) Representative images of MFC cell-derived tumors in 615-line mice (n = 5 per group). (I, J) Tumor growth curves (I) and tumor weight (J) for tumors constructed by subcutaneously co-implanting MFC cells and mouse BMDMs (Si-FADS1 or Si-NC) in 615-line mice. (K, L) Flow cytometry analysis of IFN-γ (K) and GZMB (L) in CD8⁺T cells isolated from MFC tumors. (M, N) Immunohistochemistry analysis of IFN-γ (M) and GZMB (N) in MFC tumor tissues from 615-line mice. Scale bar: 50 μm. **P < 0.01, ***P < 0.001

Based on these results, we were prompted to investigate whether FADS1⁺TAMs directly suppressed the anti-tumor immunity of CD8⁺T cells. TAMs were co-cultured with T cells derived from human PBMCs in vitro (Fig. 2D). Overexpression of FADS1 in TAMs significantly inhibited the cytotoxic functionality of CD8⁺T cells, characterized by reduced production of IFN-γ and GZMB. In contrast, knockout of FADS1 in TAMs notably reduced the suppression of CD8⁺T cell function (Fig. 2E, F). Taken together, these results suggest that FADS1 drives TAMs to acquire an immunosuppressive phenotype that effectively suppresses CD8⁺T cells.

FADS1⁺TAMs promoted GC progression by inhibiting the function of intratumoral CD8⁺T cells

To better understand the effect of FADS1⁺TAMs in GC progression, we established a mouse model of subcutaneous tumor formation by co-injecting MFC cells with mouse bone marrow-derived macrophages (BMDMs) that knocked down FADS1 (Si-FADS1) (Fig. 2G). Furthermore, through regular monitoring of tumor volume in both the control (Si-NC + MFC) and experimental groups (Si-FADS1 + MFC), we observed that tumor growth was delayed, and the tumor volume and tumor weight significantly reduced in the FADS1 knockdown group (Si-FADS1 mice) compared to the control group, indicating that FADS1⁺TAMs have an obvious tumor-promoting effect (Fig. 2H-J). Furthermore, flow cytometry confirmed that the tumor-killing function of CD8⁺T cells, including the secretion of IFN-γ and GZMB, was negatively correlated with FADS1⁺TAMs (Fig. 2K, L). IHC validated that the expression of IFN-γ and GZMB was significantly increased upon FADS1 knockdown (Fig. 2M, N), further suggesting that CD8⁺T cell-mediated anti-tumor immunity was severely disrupted by FADS1⁺TAMs. In summary, these findings indicated that FADS1⁺TAMs contributed to shaping an immunosuppressive TME, thereby facilitating GC progression.

FADS1⁺TAMs impaired anti-tumor immunity in CD8⁺T cells by inhibiting the STING-IFN-I pathway

The preceding experiments revealed that FADS1⁺TAMs contributed to shaping an immunosuppressive TME by inhibiting CD8⁺T cell function. Therefore, we investigated the mechanisms underlying this inhibition of CD8⁺T cell activity. Further analysis of the above transcriptome results revealed that overexpression of FADS1 in TAMs significantly inhibited the cytosolic DNA-sensing pathway (Fig. 3A). The cyclic GMP-AMP synthase (cGAS)- stimulator of interferon genes (STING) pathway is a key intracellular DNA sensing mechanism that triggers the secretion of type I interferons. STING signaling in immune cells drives type I IFN-dependent spontaneous T-cell priming, which enhances tumor immunogenicity and improves cancer immunotherapy [20, 21]. The results of GSEA-GO analysis revealed that the DEGs were enriched in the negative regulation and response of type I IFN (Fig. 3B). Moreover, the heatmap showed that IFN-stimulated genes (ISGs) in TAMs were downregulated following the overexpression of FADS1 (Fig. 3C).

Fig. 3.

FADS1⁺TAMs impaired anti-tumor immunity in CD8⁺T cells by inhibiting the STING-IFN-I pathway(A, B) KEGG (A) and GO (B) pathway enrichment analysis of the differential expressed genes in the transcriptome between OE-FADS1 TAMs and OE-NC TAMs. (C) Heatmap showing the expression of IFN-stimulated genes in OE-FADS1 TAMs and OE-NC TAMs. (D) Western blotting analysis of STING pathway-related proteins in TAMs (FADS1 overexpressed or knockout) after incubation with apoptotic MKN-45 cells for 6 h. (E) Real-time quantitative PCR analysis of type I IFN-response genes in TAMs (FADS1 overexpressed or knockout) after incubation with apoptotic MKN-45 cells for 24 h. (F) Western blotting analysis of STING pathway-related proteins in OE-FADS1 TAMs after incubation with apoptotic MKN-45 cells following treatment with the STING agonist diABZI (100 nM) for 6 h. (G) Real-time quantitative PCR analysis of type I IFN-response genes in OE-FADS1 TAMs after incubation with apoptotic MKN-45 cells following treatment with the STING agonist diABZI (100 nM) for 24 h. (H) Flow chart of co-culturing TAMs and T cells following treatment with the STING agonist diABZI to assess phenotypic changes in T cells (By Figdraw). (I, J) Expression profiles of IFN-γ (I) and GZMB (J) in CD8⁺T cells, quantified by flow cytometry in OE-FADS1 TAMs following treatment with the STING agonist diABZI. **P < 0.01, ***P < 0.001, ^###P < 0.001

Next, Western blot assay was performed to assess the activation of the STING pathway in TAMs following FADS1 overexpression or knockout. The total levels of intracellular STING, IRF3, and TBK1 proteins remained largely unchanged after FADS1 overexpression or knockout. However, following FADS1 overexpression, the phosphorylation of STING protein was significantly reduced, along with a decrease in the phosphorylation of downstream IRF3 and TBK1 proteins. In contrast, FADS1 knockout increased the phosphorylation levels of these proteins (Fig. 3D). Consistent with the transcriptional group results, real-time PCR showed that the expression levels of type I IFN-response genes in TAMs with FADS1 overexpression were significantly decreased, whereas TAMs with FADS1 knockout exhibited the opposite trend, with increased expression levels (Fig. 3E).

Therefore, we hypothesized that FADS1⁺TAMs may inhibit the activation of the intracellular STING pathway and suppress the expression of type I IFN genes. To test this hypothesis, we treated the co-culture system with the STING agonist diABZI. Our results showed that, compared to the control group, the STING agonist significantly reversed the inhibition of the STING pathway and restored the expression of type I IFN genes in TAMs with FADS1 overexpression (Fig. 3F, G). Additionally, we observed that the STING agonist in TAMs reversed the effects of FADS1 overexpression on CD8⁺T cell functionality, emphasizing the crucial role of the STING-IFN-I pathway in mediating FADS1⁺TAMs-induced CD8⁺T cell dysfunction (Fig. 3H- J).

FADS1⁺TAMs provides a favorable microenvironment for the immunosuppressive phenotype through efferocytosis in TME

To further explore the role of FADS1 in mediating immunosuppressive phenotype, we conducted KEGG analysis on scRNA-seq data. Our analysis revealed that genes highly expressed in FADS1⁺TAMs were primarily associated with lysosome, efferocytosis, and endocytosis pathways (Fig. 4A). We then performed gene set scoring using AUcell, which showed that FADS1⁺TAMs exhibited higher scores in the efferocytosis pathway (Fig. 4B). The TAM receptor MERTK, a tyrosine kinase, plays a crucial role in efferocytosis and the establishment of an immunosuppressive TME [22]. ScRNA-seq data revealed a significant increase in MERTK expression in FADS1⁺TAMs (Fig. 4C).

Fig. 4.

FADS1 stimulates PPARγ nuclear translocation in TAMs to promote efferocytosis and induce an immunosuppressive phenotype. (A) KEGG pathway enrichment analysis of highly expressed genes in FADS1⁺TAMs. (B) Violin plots illustrating the efferocytosis pathway scores between FADS1⁺TAMs and FADS1⁻TAMs. (C) Violin plots showing the expression of MERTK between FADS1⁺TAMs and FADS1⁻TAMs. (D) Immunohistochemistry analysis of cleaved caspase3 for tumors constructed by subcutaneously co-implanting MFC cells and mouse BMDMs (Si-FADS1 or Si-NC) in 615-line mice. Scale bar: 100 μm (left), 50 μm (right). (E) KEGG pathway enrichment analysis of differentially expressed genes between OE-FADS1 TAMs and OE-NC TAMs. (F) Representative flow cytometry plots using Annexin V/7-AAD staining to assess apoptosis in MKN-45 cells (left). Flow cytometry analysis of pHrodo intensity following co-culture of TAMs with apoptotic MKN-45 cells (right). (G) Western blotting analysis of MERTK expression levels in TAMs (OE-FADS1 and KO-FADS1). (H) Dot plots showed relative activity of highly expressed transcription factors in each cell subtypes. (I) Western blotting analysis of nuclear PPARγ protein expression levels in TAMs (OE-FADS1 and KO-FADS1). (J) Immunofluorescent images of PPAR-γ (red)/DAPI (blue) colocalization in TAMs (OE-FADS1 and KO-FADS1). Scale bar: 30 μm. (K) Western blotting analysis of nuclear PPARγ protein expression levels in TAMs (OE-FADS1 and transfected with siPPARγ). (L) Flow cytometry analysis of pHrodo intensity following co-culture of TAMs (OE-FADS1 and transfected with siPPARγ) with apoptotic MKN-45 cells. (M) Western blotting analysis of MERTK expression levels in TAMs (OE-FADS1 and transfected with siPPARγ). ***P < 0.001, ^###P < 0.001

Cancer cell apoptosis is a common event in solid tumors during malignant progression, which is cleared by TAMs through a process known as “efferocytosis” and has significant effects on immunosuppression [23]. Defects in efferocytosis contribute to enhanced anti-tumor immunity by increasing the population of immune-promoting TAMs and cytotoxic T cells within the TME [24, 25]. Research has shown that inhibiting the efferocytosis of apoptotic cells led to increased release of cyclic GMP-AMP (cGAMP) from dying tumor cells, which subsequently triggered a STING-dependent type I IFN response [24].

We further investigated whether the tumor tissues from Si-FADS1 mice contained more apoptotic cells than the control. IHC staining with cleaved caspase-3 revealed that apoptotic cells were more prevalent in the tumor tissues of Si-FADS1 mice than in the control mice (Fig. 4D). Moreover, consistent with the scRNA-seq data, in in vitro co-culture experiments, KEGG analysis revealed a strong upregulation of lysosome, endocytosis, and efferocytosis signaling pathways in OE-FADS1 TAMs (Fig. 4E).

To test whether the increased population of apoptotic cells in the TME is due to a defect in efferocytosis, we assessed the efferocytosis activity in TAMs. To do this, apoptotic cells were prepared by treating MKN-45 cells with 5-FU. When TAMs were challenged with apoptotic cells conjugated to pHrodo red dye, OE-FADS1 TAMs exhibited higher red intensity, indicating enhanced efferocytosis activity, while KO-FADS1 TAMs showed reduced red intensity, suggesting impaired efferocytosis activity (Fig. 4F). Furthermore, MERTK expression was increased in the OE-FADS1 TAMs, but was decreased in the KO-FADS1 TAMs (Fig. 4G).

Our findings suggest that the inhibition of the STING-IFN-I pathway in FADS1⁺TAMs is most likely attributed to their enhanced efferocytosis activity. Consistent with established literature, by efficiently clearing immunostimulants—specifically STING pathway agonists such as cGAMP released from apoptotic cells—macrophages prevent the overactivation of innate immune signaling. This mechanism is highly consistent with previous reports demonstrating that efferocytosis deficiency leads to the accumulation of cGAMP and subsequent pathological activation of STING [26, 27]. Further, we utilized UNC2250, the inhibitor of MERTK, to directly interrupt the efferocytosis process in OE-FADS1 TAMs. As shown in Supplementary Figure S2A, the direct blockade of efferocytosis successfully restored STING pathway activity. This further substantiates the association between enhanced efferocytosis and STING pathway suppression. Collectively, these results support the notion that FADS1 plays a critical role in the efferocytosis activity of TAMs, contributing to an immunosuppressive phenotype that fosters an adverse microenvironment.

FADS1 promoted efferocytosis activity by stimulating PPARγ nuclear translocation in TAMs

We further delved into the specific mechanism by which FADS1 regulated the efferocytosis activity in TAMs. Gene regulatory network analysis also indicated significantly elevated PPARγ activity in FADS1⁺TAMs (Fig. 4H). Western blotting revealed that the nuclear PPARγ content increased following overexpression of FADS1 and decreased following FADS1 knockout in TAMs (Fig. 4I). Immunofluorescence staining further confirmed that the nuclear localization of PPARγ was increased in OE-FADS1 TAMs, but was decreased in the KO-FADS1 TAMs (Fig. 4J).

Furthermore, we applied PPARγ small interfering RNA to knock down its expression (Fig. 4K). The results demonstrated that knocking down PPARγ decreased the efferocytosis activity in OE-FADS1 TAMs, as evidenced by reduced red intensity and MERTK expression (Fig. 4L, M). Knockdown of PPARγ reversed the effects of overexpression of FADS1 on efferocytosis, underscoring PPARγ as a pivotal mediator of FADS1-regulated efferocytosis in TAMs.

FADS1-mediated DHA synthesis promoted efferocytosis in TAMs by activating PPARγ

GSEA analysis revealed that pathways related to UFAs biosynthesis were enriched in FADS1⁺TAMs (Fig. 5A). Consistently, KEGG analysis highlighted the critical role of UFAs biosynthesis in OE-FADS1 TAMs (Fig. 5B). To explore the link between UFAs biosynthesis and efferocytosis in TAMs, we performed metabolomic profiling using liquid chromatography-mass spectrometry. ω3-poly-UFA (C22:6N3, docosahexaenoic acid [DHA]) was elevated in OE-FADS1 TAMs compared with controls (Fig. 5C). Given extensive evidence that UFAs can activate PPARγ [28], we treated KO-FADS1 TAMs with DHA, which increased the nuclear PPARγ levels and rescued efferocytosis activity (Fig. 5D-F). These findings suggest that FADS1 promotes efferocytosis in TAMs through DHA synthesis.

Fig. 5.

FADS1-mediated DHA synthesis promoted efferocytosis in TAMs by activating PPARγ. (A) GSEA plots of fatty acid biosynthetic process-related gene sets in FADS1⁺TAMs. (B) KEGG pathway enrichment analysis of the differential expressed genes in the transcriptome between OE-FADS1 TAMs and OE-NC TAMs. (C) Metabolomic profiling of differentially expressed UFAs between OE-FADS1 TAMs and OE-NC TAMs. (D) Western blotting analysis of nuclear PPARγ protein expression levels in TAMs (KO-FADS1 and treated with DHA). (E) Flow cytometry analysis of pHrodo intensity following co-culture of TAMs (KO-FADS1 and treated with DHA) with apoptotic MKN-45 cells. (F) Western blotting analysis of MERTK expression levels in TAMs (KO-FADS1 and treated with DHA). (G) Western blotting analysis of nuclear PPARγ protein expression levels in TAMs (KO-FADS1 treated with DHA and siPPARγ). (H) Flow cytometry analysis of pHrodo intensity following co-culture of TAMs (KO-FADS1 treated with DHA and siPPARγ) with apoptotic MKN-45 cells. (I) Western blotting analysis of MERTK expression levels in TAMs (KO-FADS1 treated with DHA and siPPARγ). *P < 0.05, ***P < 0.001, ^###P < 0.001

To further investigate whether FADS1-mediated DHA synthesis contributes to efferocytosis through PPARγ activation, we knocked down PPARγ in KO-FADS1 TAMs using si-PPARγ (Fig. 5G). The results revealed that the difference in efferocytosis activity between DHA-treated KO-FADS1 TAMs and controls was abolished following PPARγ knockdown (Fig. 5H, I). Taken together, these results confirmed the pivotal role of PPARγ in efferocytosis through FADS1-mediated DHA synthesis in TAMs. Based on these findings, we concluded that FADS1 activates PPARγ by synthesizing DHA, thereby promoting efferocytosis activity and contributing to an immunosuppressive phenotype.

Pharmacological targeting of FADS1 enhanced the efficacy of anti-PD-1 treatment in ectopic GC models

CD8⁺T cell infiltration and function within tumors are essential for the efficacy of immune checkpoint blockade (ICB) therapies [29]. Previously, we demonstrated that knocking down FADS1 on TAMs significantly enhanced the function of infiltrating CD8⁺T cells in GC. Based on this, we utilized the orally available, potent, and selective FADS1 inhibitor, D5D-IN-326 (Compound-326), to pharmacologically target FADS1 and validated its effectiveness in combination with anti-PD-1 treatment in GC.

Next, we established subcutaneous GC tumors in 615-line mice using MFC cells and initiated drug administration on day 3 (Fig. 6A). The results demonstrated that both the PD-1 monoclonal antibody (mAb) and the FADS1 inhibitor D5D-IN-326 significantly suppressed tumor growth, reduced tumor volume and weight, with the combination treatment showing the most pronounced tumor suppression (Fig. 6B-D). Flow cytometry analysis further revealed that treatment with either PD-1 mAb or D5D-IN-326 increased the proportions of IFN-γ⁺CD8⁺T cells and GZMB⁺CD8⁺T cells, effectively reversing immune microenvironment depletion. Notably, this effect was further enhanced by the combination of PD-1 mAb and D5D-IN-326 (Fig. 6E, F). IHC staining of the subcutaneous tumors confirmed that the combination treatment further elevated the expression of IFN-γ and GZMB, indicating that FADS1 inhibition alleviates CD8⁺T cell exhaustion (Fig. 6G, H). In summary, our findings indicate that the concurrent administration of PD-1 mAb and pharmacological targeting of FADS1 exerts a sensitizing effect, helping to reprogram the tumor immune microenvironment in preclinical ectopic models.

Fig. 6.

Pharmacological targeting of FADS1 enhanced the efficacy of anti-PD-1 treatment in GC. (A) Schematic representation of the treatment schedule for anti-PD-1 and D5D-IN-326 in 615-line mice bearing GC tumors (By Figdraw). (B) Representative images of MFC cell-derived tumors in 615-line mice (n = 5 per group). (C, D) Tumor growth curves (C) and tumor weight (D) of subcutaneously implanted MFC cells in 615-line mice after the indicated treatments. (E, F) Flow cytometry analysis of IFN-γ (E) and GZMB (F) in CD8⁺T cells isolated from MFC tumors. (G, H) Immunohistochemistry analysis of IFN-γ (G) and GZMB (H) in MFC tumor tissues from 615-line mice. Scale bar: 50 μm. *P < 0.05, **P < 0.01, ***P < 0.001

To exclude off-target effects of D5D-IN-326, we assessed the response of KO-FADS1 TAMs to the inhibitor. Notably, D5D-IN-326 treatment did not further alter efferocytosis, STING pathway activity, or CD8⁺T cell suppression in KO-FADS1 TAMs compared to vehicle controls (Supplementary Fig. S1A-C). The lack of efficacy in a FADS1-deficient background confirms that the modulatory effects of D5D-IN-326 are on-target and FADS1-dependent.

FADS1⁺TAMs as a potential exploratory indicator for tumor immunotherapy

The tumor immune microenvironment plays a critical role in the response to ICB therapy. Given the strong correlation between FADS1⁺TAMs and the immunosuppressive microenvironment, we sought to further explore the association between FADS1⁺TAMs and the response to ICB treatment. To assess the potential clinical efficacy of immunotherapy, we applied the TIDE algorithm to the TCGA-STAD cohort stratified by the FADS1⁺TAM signature. We observed that patients in the high-signature group exhibited significantly higher TIDE, immune exclusion, and T-cell dysfunction scores, alongside lower microsatellite instability (MSI) scores (Fig. 7A). Together, these results suggest that a high FADS1⁺TAM signature is closely associated with a greater capacity for tumor immune evasion and a potentially less favorable outcome following ICB treatment.

Fig. 7.

FADS1⁺TAMs as a potential exploratory indicator for tumor immunotherapy. (A) Violin plots comparing the TIDE, T-cell dysfunction, immune exclusion, and microsatellite instability (MSI) scores between the high and low FADS1⁺TAM signature groups in the TCGA-STAD cohort. (B) Schematic representation of GC patients treated with neoadjuvant chemotherapy combined with anti-PD-1 therapy in PUMCH cohort III (By Figdraw) (left). Representative CT and HE staining image of GC patients with different responses to neoadjuvant chemotherapy combined with anti-PD-1 therapy in PUMCH cohort III (right). (C) Representative images of multiplex immunofluorescence characterizing FADS1⁺TAMs in GC patients with different responses in PUMCH cohort III (red: FADS1; yellow: CD68). Scale bar: 50 μm (left). Bar chart showing the proportion of FADS1⁺TAMs among tumor cells with different responses in PUMCH cohort III (middle). ROC curves illustrating the predictive value of FADS1⁺TAMs for therapeutic response (right)

To further evaluate the potential relationship between FADS1⁺TAMs and immunotherapy outcomes in an exploratory clinical setting, tumor tissues were collected from a preliminary cohort of GC patients receiving neoadjuvant chemotherapy combined with anti-PD-1 therapy at PUMCH Hospital for multiplex immunofluorescence staining (Fig. 7B). The results indicated that non-responders exhibited a higher abundance of FADS1⁺TAMs compared to responders, and the proportion of FADS1⁺TAMs was significantly associated with patients’ treatment responses (Fig. 7C). Collectively, these exploratory findings demonstrate the potential of FADS1⁺TAMs as a preliminary indicator for ICB responsiveness in GC, warranting future biomarker-level validation.

Discussion

Emerging evidence highlights the significant role of TAMs in contributing to immunosuppressive TME and mediating therapeutic resistance, ultimately leading to a poor prognosis [30, 31]. With the advent of scRNA-seq technology, it has become widely acknowledged that the heterogeneity of TAMs is truly remarkable, closely corresponding to the state of the TME [32]. In this study, we identified a previously unrecognized association between FADS1⁺TAMs and GC progression, highlighting their potential immunosuppressive properties and the molecular mechanisms that may underlie this behavior. Furthermore, our in vivo data suggest that the inhibition of FADS1 can attenuate the immunosuppressive phenotype of TAMs, thereby exerting a sensitizing effect on immunotherapy in ectopic GC models. This highlights the promising translational potential of targeting FADS1 as a combinatorial therapeutic strategy to help overcome TAMs-associated immunosuppression in future GC treatments (Fig. 8).

Fig. 8.

Illustration of the proposed working model (By Figdraw)

Abnormal UFAs metabolism has been observed in various tumors. Evidence suggests that these metabolic alterations contribute to the immunosuppressive properties of TAMs [9]. UFAs metabolism and antitumor immunity are both disrupted during GC progression. However, the complex regulatory relationship between them remains largely unclear. In this study, using scRNA-seq data from GC, we found that the UFAs biosynthesis score in TAMs from GC tumor tissues is significantly higher than in adjacent normal tissues. Through the analysis of DEGs in TAMs from tumor and normal tissues, as well as the TCGA-STAD database, we found that FADS1 is the only UFA biosynthesis-related enzyme within TAMs that holds prognostic significance in GC. As a rate-limiting enzyme involved in the biosynthesis of long-chain PUFAs, Lee et al. revealed that FADS1 expression is up-regulated in mesenchymal-type GC cells, where it mediated the PUFAs biosynthesis pathway, playing a crucial role in ferroptosis [13]. Similarly, Morandi et al. discovered that FADS1 is highly expressed in a subset of TNBC associated with a poorer prognosis. Lipidomic analysis and functional metabolic assays revealed that FADS1-high expressing TNBC are more susceptible to ferroptosis-inducing agents [14]. Moreover, Xu et al. reported that FADS1 plays a crucial role in mediating the synthesis of arachidonic acid (AA), which regulates colorectal cancer tumor growth by promoting a high-AA microenvironment that fosters the enrichment of gram-negative microbes [15]. However, these studies primarily focused on the direct impact of FADS1 within tumor cells, overlooking a thorough exploration of its roles within the TME. In our study, by constructing a FADS1⁺TAMs signature score from bulk RNA-seq data, our clinical analysis revealed a significant negative correlation between high FADS1⁺TAM infiltration and long-term survival in GC patients. These findings suggest the role of FADS1⁺TAMs in GC progression.

Our study demonstrated that FADS1-mediated DHA synthesis activated PPARγ, thereby promoting efferocytosis in TAMs. Prior research reveals that DHA facilitates PPARγ entry into the nucleus, enhancing its transcriptional activity to promote TAMs’ M2-like differentiation [28, 33]. In colorectal cancer, tumor-secreted GRP78 entered the lipid droplets of TAMs, where it interacted with ATGL to inhibit its ubiquitination, promoting triglyceride hydrolysis and the production of AA and DHA. Excessive AA and DHA then interacted with PPARγ, activating it and mediating the M2 polarization of TAMs [34]. Moreover, in hepatocellular carcinoma, DHA released by tumor cells activates PPARγ via FABP5, leading to immunosuppressive properties in TAMs. This results in increased levels of several immune checkpoint ligands and immunosuppressive molecules, thereby impairing anti-tumor T cell immunity [10]. In the present study, we found that FADS1 mediated DHA synthesis and activated PPARγ, which in turn promoted efferocytosis by upregulating the expression of MERTK. This regulation depends on the activation of PPARγ, as silencing PPARγ abolished MERTK upregulation induced by FADS1 overexpression. PPARγ is a ligand-activated transcription factor within the nuclear receptor superfamily, which has been reported to act as an upstream regulator of MERTK [35] It is important to note that the transcriptional regulation of efferocytosis relies on a highly coordinated and context-dependent network of nuclear receptors (including PPARδ, LXRα, PPARγ, etc.), rather than a single pathway. Among these, PPARδ serves as a master regulator driving basal, physiological efferocytosis to maintain tissue homeostasis [36]. LXRα plays a critical role in specific macrophage subsets, particularly in sensing and facilitating the efflux of cholesterol derived from engulfed apoptotic cells [37]. PPARγ, while heavily involved in macrophage polarization and lipid metabolism, often functions in a highly context-specific manner during efferocytosis. Together, these receptors form an intricate network that fine-tunes the clearance of apoptotic cells depending on the tissue microenvironment and inflammatory status.

Our data suggest that activation of the STING pathway plays a crucial role in regulating anti-tumor immunity by enhancing the expression of IFN response genes following the engulfment of apoptotic tumor cells. Given that FADS1⁺TAMs exhibit strong efferocytosis activity, they efficiently clear apoptotic tumor cells, thereby reducing the accumulation of tumor-derived cGAMP and ATP, which contributes to a tumor-tolerant, immunosuppressive TME [23, 26, 27]. Some studies have shown that tumor-derived cGAMP can be released into the extracellular space within the TME, serving as a key mechanism for immune cells to activate the STING pathway [27, 38]. Inhibition of efferocytosis through treatment with blocking antibodies targeting phagocytic receptors, such as MERTK and AXL, or the disruption of LC3-associated phagocytosis in TAMs, has been shown to trigger STING-dependent type I IFN production [24, 25, 39]. Additionally, recent studies have reported that pharmacological or genetic inhibition of efferocytosis-related targets in TAMs has been shown to activate the STING-IFN-I pathway [40–42]. Consistently, we observed that STING activation was enhanced in KO-FADS1 TAMs, which have defects in clearing apoptotic cells, while STING activation was suppressed in OE-FADS1 TAMs. This suppression weakened the type I interferon response, ultimately fostering the development of an immunosuppressive TME. Moreover, it is worth noting that while metabolic products like DHA have been sporadically reported to directly modulate STING signaling, their precise role in the context of our FADS1⁺TAMs model remains to be fully elucidated [43].

CD8⁺T cells, also known as cytotoxic T lymphocytes, are pivotal in coordinating the immune response against cancer. They directly target and eliminate malignant cells by secreting effector cytokines (IFN-γ and TNF-α) and releasing cytotoxic molecules (Granzyme B and perforin) [44]. In the context of immunotherapy, especially with ICIs such as PD-1/PD-L1 blockades, the infiltration and effective activation of CD8⁺T cells are crucial for achieving clinical efficacy [29]. Nevertheless, the accumulation of exhausted CD8⁺T cells within TME is often associated with resistance to immunotherapy. In this study, we found that FADS1⁺TAMs in GC inhibited the function of CD8⁺T cells, as evidenced by reduced secretion of IFN-γ and GZMB. Pharmacological or genetic inhibition of FADS1 in TAMs reverses their immunosuppressive phenotype, enhancing their ability to activate CD8⁺T cells. This shift is reflected by an increased population of cytotoxic CD8⁺T cells within the TME. In addition, in vivo experiments using mouse ectopic GC models demonstrated that pharmacological or genetic inhibition of FADS1 markedly inhibited tumor growth, while simultaneously enhancing the tumor-killing activity of CD8⁺T cells. Notably, combining FADS1 inhibition with anti-PD-1 treatment resulted in a synergistic improvement in therapeutic efficacy in ectopic GC models. Our RNA-seq data revealed a significant correlation between FADS1⁺TAMs infiltration and the presence of exhausted CD8⁺T cells, suggesting a potential mechanism through which targeting FADS1 enhances the effectiveness of anti-PD-1 therapy. These preclinical findings suggest that targeting FADS1 in TAMs holds potential as a sensitizing strategy for ICIs therapy in GC.

Despite the promising findings presented, several limitations must be acknowledged. First, our clinical analyses are observational, while high FADS1⁺TAM infiltration strongly correlates with poor outcomes and resistance to anti-PD-1 therapy in GC patients, these correlations cannot directly establish causality. Second, our in vivo therapeutic data rely on a subcutaneous ectopic implantation model. Although appropriate for pharmacological proof-of-concept, it fails to reproduce the authentic TME- specifically the unique mucosal architecture, organ-specific stromal and immune composition, vascular context, and localized immune education. Accordingly, our conclusions must be interpreted strictly as preclinical evidence in an ectopic setting, avoiding disease-level generalizations. Finally, while systemic FADS1 inhibition (D5D-IN-326) significantly sensitized tumors to PD-1 blockade, potential pleiotropic or off-target effects across various TME compartments cannot be excluded. This pharmacological approach alone does not definitively prove that FADS1⁺TAMs are the absolute essential drivers of immune resistance. To rigorously confirm this mechanistic dependency and integrate our findings into the broader knowledge system of TAMs heterogeneity, future investigations must utilize physiologically relevant orthotopic models alongside targeted genetic approaches, such as myeloid-specific FADS1 conditional knockout mice.

Furthermore, the specific regulatory effects of FADS1 on efferocytosis within TAMs require further investigation. The transcriptional program governing efferocytosis is not a linear pathway but a coordinated network of nuclear receptors. Although we observed that PPARγ activation is associated with the upregulation of the efferocytosis receptor MERTK, our study lacks direct evidence demonstrating that PPARγ directly binds to the MERTK promoter. Therefore, we cannot exclude other potential possibilities. Future research is needed to further explore the regulatory mechanisms of PPARγ and MERTK, considering the potential synergistic roles of other nuclear receptors like PPARδ and LXRα, and to determine whether metabolite-mediated signaling and physical clearance act synergistically to suppress the STING pathway. Finally, associations regarding immunotherapy responsiveness must be interpreted strictly as exploratory. Due to the limited size of our treatment cohort and the absence of multivariable adjustments for established predictors (e.g., PD-L1, MSI, and TMB), FADS1⁺TAMs infiltration represents an exploratory indicator of therapeutic resistance. Large-scale, prospective clinical validation incorporating these standard covariates is required before this signature can be considered a definitive predictive biomarker.

In conclusion, our findings highlight the potential role of FADS1 in contributing to the immunosuppressive TME of GC. Mechanistically, our models suggest that FADS1 may enhance efferocytosis by activating PPARγ through DHA synthesis in TAMs, thereby inhibiting the STING-IFN-I pathway and impairing CD8⁺T cell-mediated anti-tumor immunity. Furthermore, high infiltration of FADS1⁺TAMs is significantly correlated with poorer therapeutic responses and clinical outcomes in GC patients. Pharmacological inhibition of FADS1 may offer a novel sensitizing strategy to enhance immunotherapy efficacy and improve outcomes for GC patients.

Supplementary Information

Below is the link to the electronic supplementary material.

Author contributions

YJ.Z., RY.W., PF.S., and YH.X. contributed equally to this work. YJ.Z., RY.W., PF.S., and YK.H. designed the study. YJ.Z., RY.W., PF.S., and YH.X. performed the majority experiments. T.Y. assisted in molecular experiments and analysis. YJ.Z., RY.W., and T.Y. analyzed the publicly available data. YJ.Z., PF.S., and T.Y collected tissue samples and clinical data from GC patients. YJ.Z., YH.X., and BW.H. conducted the statistical analyses, prepared Figures, scrutinized the results, interpreted data, and wrote/edited the manuscript. YK.H., FL.L., and JC.Y. contributed to study supervision. YK.H., JC.Y. and PF.S. provided financial support. All authors have read the final manuscript and approved the submission.

Funding

This work was supported by the National High Level Hospital Clinical Research Funding (Grant No. 2022-PUMCH-B-005), and Postdoctoral Research Funding Project of Zhejiang Province (No. ZJ2024022).

Data availability

The data supporting the findings of this study are available from the corresponding author upon request. The raw RNA sequence data reported in this paper have been deposited in the Genome Sequence Archive [45] in National Genomics Data Center [46], China National Center for Bioinformation/Beijing Institute of Genomics, Chinese Academy of Sciences (GSA-Human: HRA017221) that are publicly accessible at https://ngdc.cncb.ac.cn/gsa-human.

Declarations

Ethics approval and consent to participate

The studies involving human participants were reviewed and approved by the Institutional Review Board of Peking Union Medical College Hospital. Informed consent was obtained from all patients. Animal studies were approved by the Shanghai Yishang Biotechnology Center.

Declaration of competing interest

The authors declare that they have no competing interests.