USP1-mediated lipophagy-lipogenesis axis drives cholangiocarcinoma progression and immune evasion Authors: Yananlan Chen, Xiao Xu, Shenye Shao, Jiawei Zhang, Ziyang Wang, Changan Chen, Wangjie Jiang, Jiang Chang, Ruixiang Chen, Tao Zhou, Jifei Wang, Shuochen Liu, Yue Yu, Yaodong Zhang, Changxian Li, Xiangcheng Li Published: 2026 ABSTRACT AbstractBackgroundCholangiocarcinoma (CCA), a malignancy arising from biliary epithelial cells, features a tumor microenvironment (TME) characterized by metabolic dysregulation and immunosuppression. Although diverse metabolic aberrations have been observed, the dominant metabolic driver remains unclear. Therefore, the study aims to elucidate the molecular connections between metabolic reprogramming and immune evasion, and to provide the basis for developing TME-targeted therapies.MethodsFirst, untargeted metabolomics was performed to identify key metabolite classes involved in CCA progression. We subsequently used next-generation sequencing combined with database analysis to identify the key genes involved in tumor fatty acid metabolism. Metabolomics, proteomics and metabolic phenotyping experiments were subsequently performed to investigate this metabolic process. Tumor malignancy was evaluated using a range of in vitro and in vivo models. Single-cell RNA sequencing, Olink proteomics, flow cytometry and multiplex immunohistochemistry were performed to analyze the changes in the TME. Finally, we used humanized NOG mice to evaluate the efficacy of the combination therapy.ResultsIntegrated multi-omics analysis revealed that long-chain unsaturated fatty acids (LCUFAs) were enriched in CCA tissues and closely associated with tumor progression. Ubiquitin-specific peptidase 1 (USP1) was upregulated in CCA tissues, driving dynamic lipid metabolic reprogramming and was associated with poor prognosis. Mechanistically, USP1 orchestrated LCUFAs accumulation through coordinating the enhancement of lipophagy and lipogenesis, establishing a systems-level metabolic regulatory network. Tumor-secreted LCUFAs were preferentially internalized by tumor-associated macrophages (TAMs) through FABP5, thereby suppressing cytotoxic T cells activity. Additionally, TAMs upregulated USP1 in CCA cells, forming a positive feedback loop that perpetuates the metabolic-immune crosstalk.ConclusionIn summary, based on multi-omics analysis, this study establishes a systemic mechanism that links tumor lipid metabolism, immune suppression, and therapeutic response in CCA. Targeting the USP1-mediated metabolic-immune axis not only significantly suppresses tumor progression but also enhances the efficacy of immunotherapy. These findings construct a conceptual framework that integrates metabolic reprogramming, immune evasion, and treatment sensitivity, thereby establishing USP1 as a precision therapeutic target with broad biological and clinical importance for patients with CCA.Keywords: Immunotherapy, Tumor microenvironment - TME, Penile Cancer WHAT IS ALREADY KNOWN ON THIS TOPIC Cholangiocarcinoma (CCA) is highly aggressive and has a poor prognosis. Its microenvironment drives tumor progression and therapeutic resistance through immune editing and metabolic reprogramming. Cholangiocarcinoma (CCA) is highly aggressive and has a poor prognosis. Its microenvironment drives tumor progression and therapeutic resistance through immune editing and metabolic reprogramming. Tumor-associated macrophages (TAMs) are predominantly polarized toward the M2 immunosuppressive phenotype and can promote their own polarization by taking up long-chain unsaturated fatty acids (LCUFAs) from the tumor microenvironment (TME) via CD36. Tumor-associated macrophages (TAMs) are predominantly polarized toward the M2 immunosuppressive phenotype and can promote their own polarization by taking up long-chain unsaturated fatty acids (LCUFAs) from the tumor microenvironment (TME) via CD36. Ubiquitin-specific peptidase 1 (USP1) plays important roles in DNA repair, the cell cycle, and tumor malignant progression; however, its function in fatty acid metabolism of CCA remains unclear. Ubiquitin-specific peptidase 1 (USP1) plays important roles in DNA repair, the cell cycle, and tumor malignant progression; however, its function in fatty acid metabolism of CCA remains unclear. WHAT THIS STUDY ADDS The study reveals that USP1 stabilizes ATP-citrate lyase (ACLY) and autophagy-related 3 (ATG3) through deubiquitination in CCA, coordinately regulating fatty acid synthesis and lipophagy, thereby driving the abnormal accumulation of LCUFAs. The study reveals that USP1 stabilizes ATP-citrate lyase (ACLY) and autophagy-related 3 (ATG3) through deubiquitination in CCA, coordinately regulating fatty acid synthesis and lipophagy, thereby driving the abnormal accumulation of LCUFAs. Tumor-derived LCUFAs are found to promote M2 polarization of TAMs via the fatty acid binding protein 5 (FABP5)/peroxisome proliferator-activated receptor-γ axis, leading to the identification of an immunosuppressive FABP5+ lipid-loaded macrophage subset that impairs CD8+ T-cell function through inhibition of major histocompatibility complex-I and C-C motif chemokine ligand (CCL)4 signaling. Tumor-derived LCUFAs are found to promote M2 polarization of TAMs via the fatty acid binding protein 5 (FABP5)/peroxisome proliferator-activated receptor-γ axis, leading to the identification of an immunosuppressive FABP5+ lipid-loaded macrophage subset that impairs CD8+ T-cell function through inhibition of major histocompatibility complex-I and C-C motif chemokine ligand (CCL)4 signaling. TAMs secrete CCL3 to activate the C-C motif chemokine receptor 5/β-catenin pathway on CCA cells, leading to positive feedback upregulation of USP1 expression and establishing a “TAM-tumor” metabolic-immune crosstalk loop that perpetuates the immunosuppressive TME. TAMs secrete CCL3 to activate the C-C motif chemokine receptor 5/β-catenin pathway on CCA cells, leading to positive feedback upregulation of USP1 expression and establishing a “TAM-tumor” metabolic-immune crosstalk loop that perpetuates the immunosuppressive TME. HOW THIS STUDY MIGHT AFFECT RESEARCH, PRACTICE OR POLICY USP1 is shown to remodel the immune microenvironment via lipid metabolism regulation, offering a new avenue for investigating the function of deubiquitinases in tumor immunity. USP1 is shown to remodel the immune microenvironment via lipid metabolism regulation, offering a new avenue for investigating the function of deubiquitinases in tumor immunity. The synergistic antitumor efficacy of a USP1 inhibitor combined with macrophage depletion and anti-programmed death-ligand 1 antibody was validated in a humanized model, providing a rationale for novel clinical treatment strategies in patients with immunotherapy-resistant CCA. The synergistic antitumor efficacy of a USP1 inhibitor combined with macrophage depletion and anti-programmed death-ligand 1 antibody was validated in a humanized model, providing a rationale for novel clinical treatment strategies in patients with immunotherapy-resistant CCA. By elucidating the roles of USP1, FABP5, ACLY, and ATG3 in the tumor-TAMs interaction loop, this study offers actionable targets for developing novel small-molecule inhibitors or combination therapies. By elucidating the roles of USP1, FABP5, ACLY, and ATG3 in the tumor-TAMs interaction loop, this study offers actionable targets for developing novel small-molecule inhibitors or combination therapies. USP1 expression level and the degree of M2 macrophage infiltration are significantly associated with poor prognosis in patients with CCA, serving as potential prognostic biomarkers to guide clinical patient stratification and precision therapy. USP1 expression level and the degree of M2 macrophage infiltration are significantly associated with poor prognosis in patients with CCA, serving as potential prognostic biomarkers to guide clinical patient stratification and precision therapy. Background Cholangiocarcinoma (CCA) is characterized by aggressive progression, limited therapeutic options, and a poor prognosis.1,3 Growing evidence highlights the pivotal role of the tumor microenvironment (TME), a highly heterogeneous ecosystem that supports tumor survival and comprises diverse cellular and acellular components.4,6 The dynamic and intricate interplay between neoplastic cells and TME components facilitates tumor progression and therapeutic resistance through mechanisms such as immune editing and metabolic reprogramming.7,10 Therefore, elucidating the central regulatory role of metabolic axes within the TME is crucial for addressing the therapeutic limitations of CCA treatment. Tumor-associated macrophages (TAMs) play an important role in mediating tumor immune evasion and therapy resistance.11,13 In most solid tumors, TAMs are predominantly polarized toward the M2 phenotype, resulting in the formation of an immunosuppressive TME through the secretion of various cytokines.14 Recent studies have revealed that lipid metabolites, particularly long-chain unsaturated fatty acids (LCUFAs), also serve as key mediators of intercellular communication within the TME.15 16 Su et al reported that TAMs promote their polarization toward the M2 phenotype by upregulating cluster of differentiation 36 (CD36) expression to enable the uptake of free fatty acids (FFAs) from the microenvironment.17 Targeting the metabolic crosstalk between macrophages and tumor cells can effectively inhibit tumor metastatic progression and has clinical translational potential.18 However, in CCA, the response of TAMs to tumor-derived fatty acids, their downstream signaling transduction, and the associated immunosuppressive mechanisms remain unclear. Our previous studies revealed that hypoxia-induced upregulation of fatty acid synthesis drives tumor progression in CCA.19 However, the upstream regulatory mechanisms, particularly under normoxic conditions, remain elusive. We therefore sought to identify key regulators of aberrant fatty acid accumulation in CCA and found ubiquitin-specific peptidase 1 (USP1), a member of the ubiquitin-specific protease family. USP1 is crucial for processes such as DNA damage repair, cell cycle regulation, and stem cell maintenance.20 Previous studies have demonstrated that USP1 regulates mitochondrial dynamics, driving metabolic reprogramming and malignant progression in Hepatocellular Carcinoma (HCC).21 However, the role of USP1 in fatty acid metabolism in CCA remains unclear. This study aimed to characterize the dynamics and identify key regulators of LCUFAs in CCA. Furthermore, the mechanism by which USP1 coordinates lipogenesis and lipophagy through autophagy related 3 (ATG3) and ATP citrate lyase (ACLY) to promote fatty acid accumulation was investigated. Additionally, this study aimed to clarify the effects of LCUFAs on TAMs subsets and elucidate the TAMs-USP1 feedback loop. Finally, the potential efficacy of combination therapy will be assessed using humanized NOG mice. Materials and methods cell culture The human CCA cell lines (RBE, HCCC9810, QBC939, HuCCT1), human intrahepatic biliary epithelial cells and HEK293T cells were obtained from the National Human Genome Center at Shanghai (Shanghai, China). All cell lines were maintained in Dulbecco’s Modified Eagle Medium (DMEM; Gibco, USA) containing 10% heat-inactivated fetal bovine serum (FBS; Gibco) and 1% penicillin/streptomycin antibiotic mixture (Gibco) under standard culture conditions. Cells were incubated at 37°C in a humidified atmosphere with 5% CO2 for culture (21% O2). CCK8 assay Cell proliferation was quantitatively analyzed using the Cell Counting Kit-8 (CCK-8; Dojindo Laboratories, Kumamoto, Japan). Cells were plated in 96-well culture plates at an initial density of 1×103 cells/well. aily measurements were performed over five consecutive days by adding 10 µL of CCK-8 reagent to 100 µL of culture medium in each well. Following a 2-hour incubation at 37°C under normoxic conditions (5% CO2), absorbance was quantified at 450 nm using a multimode microplate reader (BioTek Instruments, USA). Transwell and wound‐healing assay Transwell assay Cell migration capacity was evaluated using Transwell chambers (8.0 µm pore size; Corning BioCoat Matrigel, Corning, USA). Briefly, 3×104 transfected cells suspended in 200 µL serum-free DMEM (Gibco, USA) were seeded into the upper chamber. The lower chamber was filled with DMEM supplemented with 20% FBS (Gibco) as a chemoattractant. After 48-hour incubation at 37°C under 5% CO2, non-migratory cells on the upper membrane surface were removed using cotton swabs. Migrated cells attached to the lower membrane were fixed with formaldehyde for 15 min, washed with phosphate-buffered saline (PBS), and stained with 0.1% crystal violet for 15 min. Quantitative analysis was performed by counting migrated cells in five randomly selected fields per chamber under an inverted light microscope (Olympus, Germany). Wound healing assay For wound healing assays, CCA cells were cultured to confluence in 6-well plates with serum-free medium. Standardized wounds were created using 200 µL pipette tips. Time-lapse imaging commenced post-scratch (0 hours baseline) with subsequent serum free medium exposure for 48 hours monitoring. OCR assay The extracellular oxygen consumption rate (OCR) was measured using the Extracellular OCR Plate Assay Kit (E297, Dojindo, Japan) following the manufacturer’s standardized protocol. Nile red staining To visualize lipid droplets, cultured cells were harvested and washed in 1×PBS prior to staining with Nile red solution (C2051M, Beyotime) for 15 min. Samples were then washed with PBS two times. After stained with Nile red, cells were acquired on a flow cytometer, and the data were displayed as histograms (x-axis: fluorescence intensity; y-axis: cell count). The quantitative readout for each sample was the percentage of Nile red-positive cells within the total cell population. This percentage was then normalized to that of the control group (control set as 1) to obtain the relative percentage of positive cells, which is the value plotted on the y-axis. Triglyceride and cholesterol assays Intracellular triglyceride levels were quantified using commercial assay kits (Triglyceride Content Assay Kit, Nanjing Jiancheng Bioengineering Institute, China) following the manufacturer’s standardized protocols. For data normalization, total cellular protein concentrations in lysates were determined using a bicinchoninic acid (BCA) protein quantification kit (Thermo Scientific, USA), with triglyceride values expressed as µg/mg cellular protein. Determination of ATP content The intracellular ATP content of CCA cells and TAMs was quantified using the Enhanced ATP Assay Kit (S0027, Beyotime, China) following the manufacturer’s standardized protocol. Measured values were normalized to total cellular protein content, as determined by the BCA Protein Assay Kit (Thermo Scientific, USA), with results expressed as relative luminescence units per microgram of protein. FAOBlue staining Cellular fatty acid oxidation activity was monitored using the FAOBlue fluorescent probe (FDV-0033, Japan) according to optimized protocols. Cells were washed two times with Hanks’ Balanced Salt Solution (60 min at 37℃ under 5% CO2). Real-time fluorescence imaging was conducted using a confocal microscope (ZEISS, Germany) with excitation/emission settings at 405 nm and 460 nm, respectively. Quantification was performed using a fluorescent microplate reader. RNA extraction and RT-qPCR Total RNA was isolated from cultured cells and tissue specimens using TRIzol Reagent (Thermo, USA) following the manufacturer’s guidelines. Reverse transcription was performed using HiScript Q-RT SuperMix (Vazyme, China). Quantitative PCR amplification was conducted using AceQ qPCR SYBR Green Master Mix (Vazyme, China) on a 7900HT Real-Time PCR System (Applied Biosystems, USA) under standardized cycling conditions. Melting curve analysis confirmed primer specificity. Relative messenger RNA (mRNA) expression levels were calculated using the comparative 2ˆ(−ΔΔCt) method, with β-actin serving as the endogenous normalization control. Primer sequences were designed using Primer-BLAST (NCBI) and validated for amplification efficiency. All the primer sequences were provided in online supplemental table S4. Total protein extraction and western blot analysis Total cellular proteins were extracted using NP-40 Lysis Buffer (Beyotime, China) containing 1 mM phenylmethylsulfonyl fluoride (Beyotime) for 30 min on ice, followed by centrifugation at 12,000 g for 20 min at 4℃. Protein concentrations were determined via BCA assay (Thermo, USA). Equal amounts of protein were resolved by 10% Sodium Dodecyl Sulfate-Polyacrylamide Gel Electrophoresis (SDS-PAGE) and transferred to Polyvinylidene Fluoride (PVDF) membranes (Vazyme, China). The membranes were blocked with 5% non-fat milk powder in Tris-Buffered Saline with Tween-20 (TBST) solution for 2 hours at room temperature, followed by overnight incubation with responding Horseradish Peroxidase (HRP) secondary antibody for 2 hours at room temperature. Protein bands were visualized using enhanced chemiluminescence substrate. β-actin served as the loading control for data normalization. Immunohistochemical staining and immunofluorescence staining Tissue Processing Formalin-fixed paraffin-embedded CCA tissues and mouse tumor sections were subjected to immunohistochemical staining. For antigen retrieval, slides were heated in 10 mM sodium citrate buffer (pH 6.0) at 95℃ for 15 min and cooled to room temperature. Endogenous peroxidase activity was quenched by 3% hydrogen peroxide treatment for 5 min. Immunohistochemistry Sections were blocked with 5% normal goat serum in PBS for 30 min at room temperature, followed by overnight incubation at 4℃ with primary antibodies. After PBS washes, slides were incubated with appropriate secondary antibody for 1 hour at room temperature. Diaminobenzidine was used for chromogenic development, and nuclei were counterstained with hematoxylin. Immunofluorescence Cells cultured on glass coverslips were fixed with 4% paraformaldehyde for 15 min and permeabilized with 0.1% saponin-containing immunostaining buffer (Beyotime) for 10 min. Non-specific binding was blocked with 5% bovine serum albumin in PBS for 1 hour. Primary antibody incubation was performed overnight at 4℃, followed by corresponding secondary antibody for 1 hour at room temperature. Nuclei were labeled with 4',6-diamidino-2-phenylindole (DAPI) for 5 min. Imaging and Analysis All images were acquired using a fluorescence microscope (ZEISS, Germany). Forced expression and knockdown of target genes USP1 was targeted via short hairpin RNA (shRNA) delivered through lentiviral vectors. Lentiviruses were used to construct ACLY and ATG3 overexpression cell lines. Plasmids encoding truncated proteins of USP1, ACLY and ATG3, along with mutant constructs including USP1(C90A). The cells were transfected with lipofectamine 3000 (Invitrogen, USA) according to the manufacturer’s instruction. All the sequences were provided in online supplemental table S5. Chemicals CHX: HY-12320 (50 µg/mL, MCE); MG132: HY-13259 (10 µmol/L for 6 hours, MCE); durvalumab: HY-P9919 (5 mg/kg, MCE); ML-323: HY-17543 (10 mg/kg, MCE); T0070907: HY-13202 (10 µmol/L for 24 hours, MCE); ATG7 inhibitor: HY-145371 (10 µmol/L for 8 hours, MCE); Atglistain: HY-15859 (50 µmol/L for 6 hours, MCE); Firsocostat: HY-16901 (0.1 µmol/L for 6 hours, MCE); clodronate liposomes: 40337ES08 (5 mg/10 mL, Yeasen); etomoxir: HY-50202 (20 µmol/L for 24 hours, MCE); maraviroc: HY-13004 (1 µmol/L for 24 hours, MCE); BX471: HY-12080 (1 µmol/L for 24 hours, MCE); EBSS: C0213 (Beyotime); chloroquine: HY-17589A (10 µmol/L for 24 hours, MCE). NAC: (5 mmol/L, MCE). Animal experiment Subcutaneous tumor model 6-week-old male BALB/c nude mice (GemPharmatech, China) were purchased for animal experiments. Suspensions containing 5.0×106 CCA cells were subcutaneously injected into the groin region of nude mice. The tumor was measured weekly with calipers until it was removed 4 weeks later. The tumor volume was calculated by volume=width2×length/2. Orthotopic tumor model 6-week-old male BALB/c nude mice (GemPharmatech, China) were randomly divided into five groups (n=5). A suspension containing 1×106 CCA cells was suspended in 20 µL of matrix gel and subsequently injected into the hepatic lobe of mice to establish an orthotopic model of CCA. Mouse weights were recorded every 4 days. After 4 weeks, the mice were euthanized, their livers were dissected, intrahepatic tumors were excised, and their respective weights were measured. Lung metastasis Luciferase-transfected tumor cells (5×105) were intravenously injected into the tail vein of 6-week-old BALB/c nude mice (GemPharmatech, China) in 100 µL sterile PBS. 4 weeks post-injection, mice were intraperitoneally injected with D-luciferin (GoldBio, USA). Bioluminescence imaging was performed 10 min later using an IVIS 100 Imaging System (Xenogen, Hopkinton, Massachusetts, USA). Following imaging, lungs were excised, fixed in 4% paraformaldehyde for 24 hours, and paraffin-embedded. Serial 4 µm sections were stained with H&E staining for metastatic nodule quantification. Spontaneous tumor model 6-week-old male C57BL/6 mice (GemPharmatech, China) were randomly allocated into two experimental groups (n=5 per group). To establish the USP1 knockdown model, mice received tail vein injections of either AAV-TBG-NC (control) or AAV-TBG-shUSP1 (knockdown) vectors, followed by a 4-week expression period. For induction of Intrahepatic Cholangiocarcinoma (ICC), mice subsequently underwent hydrodynamic tail vein injection at a 20 µg:5 µg ratio of transposon to transposase-encoding plasmid (YAP/AKT/NICD models). Humanized NOG mouse model NOG (NOD-scid IL2Rgnull) mice (5-week-old females, average weight: 15.6 g Shanghai Charles River) were established by tail vein injection of 1×10⁷ healthy donor-derived peripheral blood mononuclear cells (PBMCs) and simultaneous subcutaneous implantation of 1×10⁷ CCA cells into the right axilla. Blood samples were obtained from the orbit of the mice, and the proportion of human CD45-positive PBMCs was determined by flow cytometry by using CD45 antibodies (BioLegend, USA) once a week. When tumor volume reached 100 mm³, mice were randomized into experimental and control groups (n=5/group). The experimental group received intraperitoneal injections of anti-programmed cell death-ligand 1 (PD-L1) antibody (5 mg/kg), USP1 inhibitor (10 mg/kg), and clodronate liposomes (200 µL) every 48 hours for 2 weeks, while controls received equivalent doses of IgG and Dimethyl Sulfoxide (DMSO). Record the weight of the mice every week. Tumor size was measured every 3 days using digital calipers until euthanasia at day 30, after which tumors were excised for weight analysis and immunohistochemistry (IHC). All procedures were approved by the Nanjing Medical University Animal Ethics Committee and conducted in compliance with ARRIVE guidelines. Bioinformatic analysis Public datasets of mRNA expression data in CCA tissues were analyzed from The Cancer Genome Atlas (TCGA), GSE107943, GSE32879, GSE138709 and GSE171899 database. Criteria in TCGA, GSE32879 and our tissues was set as: log2fold change (FC) >0.5, p<0.01. Human tissue samples and microarray Tumor tissues and matched paratumoral specimens were obtained from 111 patients with CCA undergoing curative resection at the First Affiliated Hospital of Nanjing Medical University between 2006 and 2017. USP1 expression was evaluated through a semi-quantitative immunohistochemical scoring system, where staining intensity (0: negative; 1: weak; 2: moderate; 3: strong) was multiplied by the percentage of positive cells (0: 0–5%; 1: 6–35%; 2: 36–70%; 3: >70%) to generate final immunoreactivity scores. Two blinded pathologists independently scored each specimen, with discrepancies resolved by consensus review. USP1 expression was classified as upregulated (score >4) or downregulated (score ≤4) based on established thresholds. All patients were prospectively followed until death or October 2019, with written informed consent obtained under protocols approved by the Institutional Review Board of Nanjing Medical University Affiliated Hospital in accordance with the Declaration of Helsinki. Single-cell RNA sequencing Preparation and Quality Control of Single-Cell Suspensions Single-cell suspensions were prepared from fresh tissues using enzymatic/mechanical dissociation protocols optimized for specific sample types. Critical quality parameters were rigorously validated: total cell count >5,00,000, viability >85% (assessed by Trypan Blue exclusion), concentration 700–1,200 cells/µL in ≥100 µL volume, cell diameter <40 µm, aggregate rate <15%, minimal debris, and absence of reverse transcription inhibitors or exogenous nucleic acids. Qualified suspensions were immediately processed for 10x Genomics single-cell transcriptomic workflows. Single-Cell Library Construction Single cells, gel beads containing cell barcodes (10x Genomics Chromium Next GEM Chip), and reagents were co-encapsulated into Gel Bead-in-EMulsions (GEMs) via microfluidic partitioning. Within GEMs, cells were lysed to release RNA, which hybridized to poly(dT) primers containing cell barcodes and unique molecular identifiers (UMIs). Reverse transcription was initiated using template switch oligos with rGrGrG sequences complementary to CCC-extended complementary DNA (cDNA) ends. Post-reaction, GEMs were disrupted and cDNA products purified using SPRIselect magnetic beads (Beckman Coulter). cDNA Amplification and Quality Assessment Full-length cDNA was amplified via 12-cycle PCR (Veriti Thermal Cycler, Thermo Fisher Scientific), followed by purification. cDNA quantity was measured using Qubit V.4.0 Fluorometer (Thermo Fisher Scientific), with integrity verified via Agilent 2100 Bioanalyzer High Sensitivity DNA assay (Agilent Technologies). Samples meeting thresholds (>1.8 ng/µL concentration, DV200 >60%) proceeded to library preparation. Sequencing Library Construction and Sequencing Libraries were generated through enzymatic fragmentation, end repair/A-tailing, adapter ligation, and index PCR using Chromium Single Cell 3ʹ Library Kit (v3.1). Final libraries were quantified (Qubit V.4.0) and size-distribution validated (Qseq400 Bio-Fragment Analyzer, BiOptic). Sequencing was performed on Illumina NovaSeq 6000 with paired-end 150 bp reads, targeting 20,000–50,000 reads per cell at >85% Q30 bases. Data Analysis Pipeline Raw data were processed using Cell Ranger (V.7.1.0, 10x Genomics) for demultiplexing, alignment (GRCh38 reference genome), and UMI counting. Downstream analyses in Seurat (V.5.0.1) included: First, normalize gene expression levels to correct for sequencing depth and reduce technical noise. Then, identify highly variable genes (those with high cell-to-cell variation) to improve the accuracy of cell clustering (default: 3,000 genes). Subsequently, multiple samples were merged using a combination of canonical correspondence analysisand nearest neighbors algorithms to correct for batch effects. Although each droplet is ideally expected to encapsulate a single cell, experimental artifacts such as empty droplets, doublets, or multiple cells can occur. In addition, dying or lysed cells often exhibit elevated mitochondrial gene expression. Therefore, cell quality control was performed based on three key metrics: total UMI count per cell, mitochondrial gene percentage, and number of detected genes. The following thresholds were applied: (1) mitochondrial gene percentage ≤20%: High values may indicate apoptosis or cell damage; (2) minimum number of UMIs per cell ≥100: ensuring sufficient transcript capture; (3) gene count per cell between 500 and 7,000: this range helps exclude doublets, multiple cells, or compromised cells while retaining genuine single-cell profiles. (4) Filter cells expressing more than 7,000 genes, as gene counts exceeding 7,000 may indicate doublets or multiplets. Re-normalize the expression values, then perform Principal Component Analysis (PCA) to reduce dimensionality. Next, apply SNN (shared nearest neighbor) clustering to group cells based on similarity and identify distinct subpopulations. Then, visualize the reduced-dimensional data using t-Distributed Stochastic Neighbor Embedding (t-SNE) and Uniform Manifold Approximation and Projection (UMAP). Cells with similar gene expression patterns will appear closer together in the resulting t-SNE or UMAP plots. Finally, differentially expressed genes identification (Wilcoxon rank-sum test) and cell–cell communication network prediction were performed (CellChat). The raw data have been deposited in the China National Center for Bioinformation (CNCB) database under the accession codes subCRA039320. IP coupled with mass spectrometry Total cellular proteins were isolated from CCA cells using Radio Immunoprecipitation Assay (RIPA) lysis buffer supplemented with protease inhibitors. Immunoprecipitation was conducted overnight at 4°C with 2 µg of anti-USP1 monoclonal antibody (14346–1 AP, Proteintech, China) followed by incubation with Protein A/G-agarose beads (Thermo Fisher Scientific, Waltham, Massachusetts, USA) for 4 hours. The mass spectrometry analyses were carried out by BGI Tech Solutions (BGI Shenzhen, Guangdong, China), using a Q-Exactive HF X tandem mass spectrometer (Thermo Fisher Scientific, San Jose, California, USA). GST-Pull down Glutathione S-transferase (GST)-USP1 were cloned into the pGEX-4T-1 vector and then transformed into Escherichia coli BL21 (DE3, C1400, Solarbio). The IPTG (I8070, Solarbio) was used to induce the expression of GST-USP1 protein at 16℃ with 0.5 mM. The bacteria were harvested and purified with GST-Tag Protein Purification Kit with Magnetic Agarose Beads (P2260S, Beyotime Biotechnology) according to the manufacturer’s protocol. The purified beads were eluted and subsequently incubated for overnight at 4℃ with lysates from HEK 293T cells overexpressing either Myc-ATG3 or Flag-ACLY, followed by western blot analysis. Macrophages from peripheral blood of healthy donors Human peripheral blood-derived macrophages were generated through the following standardized procedure: Peripheral blood samples from healthy donors were acquired with Institutional Review Board approval. Mononuclear cells were isolated using density gradient centrifugation, followed by monocyte enrichment with the MojoSort Human Pan Monocyte Isolation Kit (#480059, BioLegend). Purified monocytes were subsequently cultured for 7 days in medium containing 100 ng/mL macrophage colony-stimulating factor (#574804, BioLegend) to induce macrophage differentiation. Transmission eletron microscopy Transmission electron microscopy (TEM) analysis of autophagosomes was performed following standardized ultrastructural preparation protocols. Cellular samples were primarily fixed in 2.5% glutaraldehyde for 30 min at room temperature, followed by embedding in pyramid-tip molds through sequential ethanol dehydration and 72 hours thermal polymerization at 60℃. Ultrathin sections (100 nm thickness) were mounted on copper grids and dual-stained with 1% uranyl acetate (20 min) and Reynolds’ lead citrate (10 min) under ambient conditions. TEM imaging was conducted using a Hitachi H-7650 system at 80 kV accelerating voltage, with digital micrographs acquired through the AMT-600 Image Capture Engine software. mRFP-GFP-LC3 puncta assay Adenoviral vectors encoding mRFP-GFP-LC3 (GenePharma, Shanghai) were introduced into appropriately seeded 24-well plates through 72-hour infection. Cellular fluorescence patterns were analyzed via LSM880 confocal microscopy (Zeiss). The mRFP signal persistently marked LC3 localization, while GFP attenuation reflected autophagosome-lysosome fusion (autolysosome formation) and subsequent acidic pH-induced fluorescence quenching. Post-image acquisition, merged fluorescence micrographs enabled differential quantification: yellow puncta (mRFP+GFP+) represented autophagosomes, whereas red puncta (mRFP+GFP−) indicated autolysosomes. Autophagic flux intensity was calculated based on puncta density ratios. Enzyme-linked immunosorbent assay The content of C-C motif chemokine ligand 3 (CCL3) and interferon (IFN)-γ in intracellular and supernatant was quantified using the CCL3 kit (Bioswamp) and the IFN-γ kit (Bioswamp). Flow cytometry Tumor specimens underwent mechanical dissection into 1 mm³ explants followed by PBS rinsing. For tissue dissociation, 50 µg/mL DNase I (solarbio, D8072) and 2 mg/mL collagenase IV (solarbio, C8160) were added to the tumor solution and incubated in a 37°C incubator for 40 min. And then collect cells by sequential filtration through 70 µm mesh. Centrifugal pelleting (500×g, 10 min) yielded cellular precipitates for flow cytometry preparation (500 µL resuspension buffer). The products used are as follows: CD68 (PE): BioLegend, 333807; CD45 (FITC): Bioss, bsm-30095M-FITC; CD45 (FITC): Bioss, bsm-30251A-FITC; CD11b (PE/Cy7): Bioss, bsm-30073M-PE-Cy7; CD11b (PE/Cy7): BioLegend, 101215; CD206 (APC): eBioscience, 2473804; CD206 (APC): eBioscience, 141707; CD80 (PE): eBioscience, 2356227; F4/80 (FITC): BioLegend, 123107. The flow cytometry results were analyzed using FlowJo V.10 software. Reactive oxygen species assessment Intracellular oxidative stress was quantified using a commercial reactive oxygen species (ROS) detection system (Yeasen Biotechnology, Shanghai). Transfected cells were incubated with 10 µM DCFH-DA fluorescent probe in serum-free medium (37°C, 30 min, light-protected). Following PBS washing cycles, redox-sensitive fluorescence signals were visualized through fluorescence microscopy (Nikon Eclipse Ti, 488 nm excitation). Fatty acid uptake and FITC-OA Fatty acid uptake The fatty acid uptake capacity of CCA cells and tissue microarrays (TMAs) was measured by Screen Quest Fluorimetric Fatty Acid Uptake Assay Kit (36385, AAT Bioquest). Following the manufacturer’s standardized protocol. The specific protocol follows the manufacturer’s standardized protocol. FITC-OA Tumor cells were incubated with FITC-OA (Q-0087819, XI’AN Qiyue Biology) for 24 hours, then the supernatant of tumor cells was collected to treat macrophages for 48 hours. Luciferase reporter assay Three copies of a peroxisome proliferator-activated receptor-γ (PPARγ) consensus binding site were inserted into the pGL4 vector to construct a PPARγ luciferase reporter vector. For transcriptional activity assessment, macrophages were co-transfected with 2 µg of the pGL4 plasmid and 1 µg of pRL-TK normalization plasmid (Promega). Post-transfection cellular lysates were analyzed using the Dual-Luciferase Reporter Assay System (Cat# E1910, Promega) according to manufacturer specifications. Signal quantification was performed with a Fluorescent Microplate Reader. T cells activation and proliferation assays CD8+ T cells were isolated from PBMCs of healthy donors through EasyIso Human CD8 T Cell Isolation Kits (Eynos LifeScience, SN3103). Subsequently, CD8+ T cells were labeled with CFSE (C34570, Thermo Fisher Scientific) and cultured for 3 days at 37°C with 2 µL of αCD3/28 beads (Miltenyi Biotec), 30 U recombinant IL-2 (212–12, PeproTech) for 3 days. Flow cytometry detected CD8+ T cells proliferation level. RNA sequencing Total RNA was extracted from macrophages or tumor tissues using TRIzol reagent (Invitrogen, California, USA). The purity and concentration of total RNA were measured by a NanoDrop 2000 nucleic acid protein analyzer (Termo Scientific, Massachusetts, USA). RNA sequencing analysis was performed by Biomarker Technologies (Beijing, China). A total of 1 µg RNA per sample was used for library construction. Sequencing libraries were prepared using the Hieff NGS Ultima Dual-mode mRNA Library Prep Kit for Illumina (Yeasen Biotechnology, Shanghai) according to the manufacturer’s protocol, with unique index codes assigned to each sample. Briefly, mRNA was isolated from total RNA using poly-T oligo-attached magnetic beads. First-strand cDNA was synthesized, followed by second-strand synthesis. The resulting DNA fragments were blunt-ended using exonuclease/polymerase, and 3′ ends were adenylated. NEBNext adaptors with hairpin loops were ligated to the fragments. After purification with AMPure XP beads (Beckman Coulter), the adaptor-ligated cDNA was treated with USER Enzyme (3 µL; NEB) at 37°C for 15 min and then at 95°C for 5 min. PCR amplification was carried out using Phusion High-Fidelity DNA Polymerase with universal and index primers. The PCR products were purified with AMPure XP beads, and library quality was evaluated using the Agilent Bioanalyzer 2100 system. Finally, the libraries were sequenced on an Illumina NovaSeq platform to generate 150 bp paired-end reads. The raw data have been deposited in the CNCB database under the accession codes OMIX012009. Untargeted metabolomics analysis Untargeted metabolomics analysis was performed by Cosmos Wisdom Technologies (Hangzhou, China). The data have been deposited in the CNCB under the project accession numbers OMIX012029. Sample Preparation 30 pairs of CCA tissues and paired normal bile duct epithelial tissues were collected. Each sample was homogenized with 1 mL of pre-chilled methanol-water (4:1, v/v) and 3 mm zirconium oxide beads using a tissue homogenizer at 60 Hz for 60 s, followed by ultrasonication (200 W, 10 min). After incubation at 4°C for 30 min, the samples were centrifuged at 12,000 g for 15 min (4°C). Mix an equal amount of supernatant from each sample to create a quality control (PQC) sample. After vacuum freeze-drying, resuspend in 100 µL of 10% methanol/90% water solution, vortex for 30 s, sonicate for 1 min, centrifuge at 14,000 g, 4°C for 10 min, transfer the supernatant to a sample vial for mass spectrometry analysis. LC-MS Analysis Metabolites were separated using a Waters ACQUITY BEH C18 Column (17 µm 2.1 mm × 100 mm) on a Vanguish Flex UPLC equipped with a refrigerated autosampler (10°C) and column heater (40°C). Two mobile phase condition was used to improve the metabolite coverage. Condition 1: Solvent A (0.1% formic acid in water) and solvent B (0.1% formic acid in acetonitrile/methanol=4/6) were used to elute the metabolites with a 13.5-minute gradient, as follows: 10% B at 0 min, 0.25 mL/min; 40% B at 3 min, 0.25 mL/min; 95% B at 5 min, 0.25 mL/min; 100% B at 8 min, 0.6 mL/min; 100% B at 10 min, 0.6 mL/min; and back to 10% B at 10.5 min, 0.25 mL/min; and equilibrate for 3 min. Samples were analyzed using a Q Exactive HF-X (QE-HF-X) mass spectrometer equipped with a heated electro-spray ionization source. All the data were acquired in positive and negative switching mode using full scan detection, and the PQC were also analyzed with full scan/ddMS2 to acquire MS2 fragmentation for metabolite identification and annotation. Condition 2: Solvent A was 6.5 mM NH4HCO3 in water, solvent B was 6.5 mM NH4HCO3 in methanol. The 12-minute gradient was employed: 0 min, 0.25 mL/min, 10% B; 4 min, 0.25 mL/min, 40% B; 6 min, 0.25 mL/min, 95% B; 8 min, 0.4 mL/min, 99% B; 8.5 min, 0.4 mL/min, 99% B; 8.6 min, 0.25 mL/min, 10% B; 12 min, 0.25 mL/min, 10% B. Data acquisition were performed only in negative mode, and full scan was used for all the samples, and full scan/ddlS2 was also used for PQC samples. Metabolite detection and identification were carried out using MS-DIAL (V.5.1.230912) by querying online databases (MoNA, GNPS, HMDB, and MS-DIAL database) and an in-house library. Retention time drift was corrected by aligning all samples to the PQC reference with a tolerance of 0.1 min. Retention time drift correction was performed by aligning all the samples to the reference file (PQC) with tolerance of 0.1 min. Differential Metabolite Analysis Based on the log2 transformed raw data, differential metabolites are identified using both univariate and multivariate statistical analysis. Initially, the t-test function in R is employed for univariate analysis to test for significant differences in means between two groups of samples (p value<0.05). Subsequently, multivariate analysis (orthogonal partial least square-discriminant analysis, OPLS-DA) is carried out using the ropls R package with orthoI=NA set as the default parameter. The variable importance in projection (VIP) obtained from the OPLS-DA model is used to preliminarily screen differential metabolites between different samples or groups (default threshold: VIP >1). In addition, the p value/false discovery rate values from univariate analysis further aid in the selection of differential metabolites. Generally, variables meeting both criteria of p value<0.05 and VIP >1.0 are considered as differential metabolites. Lipidomics analysis When NC and Sh-USP1 CCA cells reached 70–90% confluence in culture dishes, the following procedures were performed: (1) The original medium was aspirated and cells were gently washed three times with PBS; (2) cells were detached using 0.25% trypsin-EDTA solution (with phenol red indicator) at 37°C, with digestion terminated by complete medium containing 10% FBS on observing cell rounding under microscopy; (3) cell pellets were collected by centrifugation at 1,100 rpm for 5 min followed by three PBS washes to remove residual medium components; (4) after supernatant removal, cell pellets were rapidly transferred to pre-chilled cryovials, flash-frozen in liquid nitrogen (<2 min). Subsequently, lipidomics analysis was performed by Biomarker Technologies (Beijing, China). The data have been deposited in the CNCB under the project accession numbers OMIX009595. Metabolites Extraction The liquid chromatography coupled with mass spectrometry (LC-MS) system used for metabolomics analysis consisted of Waters Acquity I-Class PLUS ultra-high performance liquid chromatography system coupled with a Waters Xevo G2-XS QTof high-resolution mass spectrometer. Separation was performed using a Waters Acquity UPLC HSS T3 column (1.8 µm 2.1 × 100 mm). Positive ion mode: mobile phase A: 0.1% formic acid aqueous solution; mobile phase B: 0.1% formic acid acetonitrile. Negative ion mode: mobile phase A: 0.1% formic acid aqueous solution; mobile phase B: 0.1% formic acid acetonitrile. The injection volume was 1 µL. LC-MS/MS Analysis The Waters Xevo G2-XS QTOF high-resolution mass spectrometer was operated in MSe mode to simultaneously acquire both primary and secondary mass spectrometry data, controlled by MassLynx V.4.2 (Waters). Each acquisition included parallel data collection at low collision energy (2V) and high collision energy (10~40V), with a scan rate of 0.2 s per spectrum. The parameters of the ESI ion source are as follows: Capillary voltage: 2,000V (positive ion mode) or −1,500V (negative ion mode); cone voltage: 30V; ion source temperature: 150°C; desolvent gas temperature: 500°C; backflush gas flow rate: 50 L/hour; desolventizing gas flow rate: 800 L/hour. Data Analysis Peak area data were normalized against the total peak area prior to subsequent analysis. Principal component analysis and Spearman correlation analysis were used to assess repeatability within sample groups and the quality control samples. Differential analysis was conducted based on group information, including fold-change calculation and independent t-tests to determine significance (p value). OPLS-DA modeling was performed using the R package ropls, with model validity assessed through 200 permutation tests. VIP values were derived from cross-validation. Differential metabolites were selected based on the combined criteria of p value<0.05 and VIP >1. Onlink target 96 Cell lysates were collected from normal CCA cells and CCA cells co-cultured with TAMs for 48 hours. Cytokine profiling in co-cultured CCA cells was analyzed using the Onlink 96-target protein detection array. Prior to sequencing using the (Illumina NovaSeq 6000, Hangzhou Cosmos Wisdom Biotech, Hangzhou, China), libraries go through a bead-based purification step and the quality is assessed using the Agilent 2100 Bioanalyzer (Agilent Technologies, Palo Alto, California, USA). The raw output data is quality controlled, normalized and converted into Normalized Protein eXpression, Olink’s proprietary unit of relative abundance. Data normalization is performed using an internal extension control and an external plate control to adjust for intra-run and inter-run variation. All assay validation data (detection limits, intra-assay and inter-assay precision data, predefined values, etc) are available on the manufacturer’s website (www.olink.com). The raw data have been deposited in the CNCB database under the accession code OMIX009596. Statistical analysis Statistical analyses were conducted using SPSS V.26.0 (IBM, Armonk, New York, USA) and GraphPad Prism V.7 (GraphPad Software, San Diego, California, USA). Continuous variables were compared by Student’s t-test, while categorical associations were analyzed with χ² tests. Survival curves were generated through Kaplan-Meier analysis with log-rank testing. Multivariate Cox regression models assessed independent prognostic factors. Correlation analyses employed Pearson’s coefficients. Statistical significance was defined as *p<0.05, **p<0.01, and ***p<0.001. Results Multi-omics sequencing revealed abnormal lipid metabolism in human CCA microenvironment Metabolic reprogramming is a characteristic of the TME. To elucidate metabolic dysregulation in CCA, nontargeted metabolomics analysis was conducted using LC-MS/MS to identify differentially abundant metabolites between 30 pairs of CCA and their paired adjacent normal tissues (figure 1A). Compared with those in adjacent normal tissues, 161 metabolites were altered in CCA tissues, of which 90 metabolites were upregulated and 71 were downregulated (figure 1B and C). The metabolites exhibiting the most significant differences were predominantly lipids (figure 1D). Moreover, we observed that a substantial number of differentially abundant metabolites were distributed in the lipid metabolism pathway (figure 1E). Metabolomic pathway analysis (MetPA, https://www.metaboanalyst.ca/) using 161 differentially abundant metabolites revealed that the biosynthesis of primary bile acid, unsaturated fatty acids, the metabolism of glycerophospholipid, taurine, hypotaurine, alpha-linolenic and glycerolipid were enriched in CCA tissues (figure 1F and online supplemental figure S1A). Furthermore, metabolite set enrichment analysis revealed that the glycerophospholipid metabolism pathway was the most significantly upregulated differentially expressed pathway identified (online supplemental figure S1B). Previous studies have confirmed that LCUFAs play an important role in tumor progression, though this role remains unclear in CCA.22 23 According to the metabolic analysis, oleic acid (C18:1, OA) was the most abundant LCUFA in tumor tissues (figure 1G). Further study revealed that OA expression was greater in normal tissue than in well-differentiated adenocarcinoma tissues. Moreover, the content of OA in CCA tissues increased with tumor progression (figure 1H). In addition, the OA content was significantly correlated with lymph node metastasis, liver invasion and tumor size (figure 1I–J). To investigate the regulatory genes involved in aberrant lipid metabolism in CCA, RNA sequencing was performed on 11 of the 30 CCA tissues with high levels of LCUFAs and their paired adjacent normal tissues (figure 1K). Kyoto Encyclopedia of Genes and Genomes (KEGG) and Gene Set Enrichment Analysis (GSEA) enrichment analyses revealed that the differentially genes were enriched in fatty acid metabolism, the PPAR signaling pathway and the fatty acid degradation pathway (figure 1L and M). Gene Ontology (GO) enrichment analysis demonstrated that many lipid metabolism-related biological processes were significantly altered in CCA tissues compared with paired adjacent normal tissues (online supplemental figure S1C). Integrative transcriptomics and metabolomics analyses revealed that fatty acid degradation was the pathway associated with the most significant difference (online supplemental figure S1D). These findings demonstrate that LUCFAs play a significant role in CCA progression. However, the precise molecular determinants and regulatory mechanism underlying the metabolic dysregulation of LUCFAs require systematic elucidation. Multi-omics sequencing revealed abnormal lipid metabolism in Human CCA microenvironment. (A–C) Non-targeted metabolomic sequencing was performed on 30 pairs of CCA tissues and matched adjacent normal bile duct epithelial tissues. Differential analysis (criteria: p<0.05, VIP>1) revealed 161 differentially abundant metabolites, including 90 upregulated and 71 downregulated metabolites. (D) The top 30 most significantly differentially abundant metabolites are displayed, primarily concentrated in lipid metabolism pathways. (E) The distribution of the identified differentially abundant metabolites. (F) Pathway enrichment analysis of differentially abundant metabolites revealed upregulated lipid-related metabolic pathways. (G) LCUFAs accumulated in CCA tissues, with OA exhibiting the greatest accumulation abundance. (H) OA was significantly downregulated in the early stages of CCA development but markedly upregulated in the middle-to-late stages. (I–J) OA levels were significantly correlated with lymph node metastasis (p=0.004, effect size (r): 0.514, 95% CI 0.405 to 1.884), liver invasion (p=0.037, effect size (r): 0.382, 95% CI 0.054 to 1.666), and tumor size (p=0.014, effect size (r): 0.443, 95% CI 0.099 to 0.693) in CCA. (K) Transcriptomic sequencing of 11 pairs of CCA and matched adjacent tissues revealed 1,244 DEGs (screening criteria: p<0.05 and |log2FC|>1), including 936 upregulated genes and 308 downregulated genes. (L–M) Pathway enrichment analysis of differentially expressed genes revealed the most significant alterations in lipid metabolism-related pathways, including the PPAR signaling pathway, fatty acid metabolism, and fatty acid degradation. *p<0.05, **p<0.01, ***p<0.001. CCA, cholangiocarcinoma; CCK, Cell Counting Kit; DEG, differentially expressed gene; FC, fold change; GSEA, Gene Set Enrichment Analysis; KEGG, Kyoto Encyclopedia of Genes and Genomes; LC-MS, liquid chromatography coupled with mass spectrometry; LCUFAs, long-chain unsaturated fatty acids; OA, oleic acid; OPLS-DA, orthogonal partial least square-discriminant analysis; PPAR, peroxisome proliferator activated receptor; VIP, variable importance in projection. USP1 was highly expressed in CCA and reprogrammed fatty acid metabolism We subsequently combined the transcriptome sequencing results and data from TCGA and GSE32879 databases to screen key genes related to fatty acid metabolism and identified USP1 and EPHB2 as candidate genes. The criteria for the TCGA and GSE32879 data and data for our sample tissues were as follows: log2FC >0.5 and p<0.01. Genes associated with fatty acid metabolism were screened from the GeneCards database on the basis of a predefined relevance score cut-off of >10 (figure 2A). We performed GSEA enrichment analysis by dividing the samples into groups based on the median expression levels of USP1 and EPHB2 in TCGA. The results showed that USP1 was significantly associated with multiple lipid metabolism pathways, such as fatty acid biosynthetic, lipid binding, lipid biosynthetic and lipid metabolic, whereas no significant enrichment in lipid metabolism pathways was detected for EPHB2 (online supplemental figure S2A). Given that, we selected USP1 as the focus for subsequent research. To further validate the oncogenic role of USP1, we analyzed USP1 expression on the basis of TCGA pancancer transcriptomics data (http://gepia2.cancer-pku.cn/#index) and found that USP1 was highly expressed in a variety of tumors (online supplemental figure S2B). We also found that USP1 was highly expressed in CCA tissues in the GSE107943 dataset (online supplemental figure S2C). To define the specific role of USP1 in CCA, we assessed the mRNA and protein levels of USP1 in CCA and paired adjacent normal tissues from our center and observed a significant increase in the expression of USP1 in CCA tissues (figure 2B and online supplemental figure S2D). Upregulation of USP1 expression was similarly observed in CCA cells relative to normal biliary epithelial cells (online supplemental figure S2E). To further analyze gene expression in CCA, we performed IHC staining on a TMA derived from 111 patients with CCA and observed significantly increased expression of USP1 in CCA tissues (figure 2C and online supplemental figure S2F). Kaplan-Meier survival analysis, which was based on the TMA results, revealed that patients with high USP1 expression had a significantly worse prognosis (figure 2D and online supplemental table S1). Univariate and multivariate analyses revealed that USP1 expression was an independent predictor of postoperative overall survival (OS) (table 1). USP1 was highly expressed in CCA and reprogrammed fatty acid metabolisms. (A) Fatty acid metabolism-related dysregulated genes were screened by integrating the TCGA database, GSE32879 database, in-house sequencing results (p<0.01, log2FC >0.5), and fatty acid metabolism-related genes from GeneCards (Relevance >10). (B) Western blot analysis of USP1 expression was performed in six pairs of CCA tissues and matched adjacent normal tissues. The results showed that USP1 was highly expressed in CCA tissues. (C) IHC staining of USP1 in a tissue microarray containing 111 paired carcinoma and adjacent tissues from patients with CCA demonstrated significant upregulation of USP1 expression in tumor tissues. (D) Survival analysis revealed that USP1 upregulation correlated with poor prognosis in patients with CCA. (E–G) Untargeted lipid metabolomics profiling of control and USP1-knockdown CCA cells (p<0.05, VIP >1) revealed 12 upregulated and 92 downregulated lipid species. (H) Enrichment analysis revealed that these differential lipids were predominantly enriched in glycerophospholipid metabolism, unsaturated fatty acid biosynthesis, fatty acid degradation, fatty acid synthesis, and fatty acid elongation pathways. (I–J) According to untargeted lipid metabolomics, USP1 knockdown significantly altered the cellular levels of LCUFAs in CCA cells, with OA showing the most pronounced change, while the TG content remained unaffected. (K) FFA levels were quantified in USP1-knockdown or USP1-overexpressing CCA cells confirmed that USP1 knockdown or overexpression significantly modified FFA levels. (L–M) TG quantification (L) and Nile red staining (M) were performed to explore the influence of USP1 on LD accumulation. USP1 modulation did not influence lipid droplet accumulation. (N–P) Extracellular OCR measurements (N), FAO blue staining (O), and ATP content detection (P) collectively indicated that USP1 knockdown substantially impaired FAO capacity. *p<0.05, **p<0.01, ***p<0.001. CCA, cholangiocarcinoma; FAO, fatty acid oxidation; FC, fold change; FFA, free fatty acid; IHC: immunohistochemical; LCUFAs, long-chain unsaturated fatty acids; LD, lipid droplet; OA, oleic acid; OCR, oxygen consumption rate; RNA-seq, RNA sequencing; TCGA, The Cancer Genome Atlas; TG, triglyceride; USP1, ubiquitin-specific peptidase 1; VIP, variable importance in projection. Table 1Univariate and multivariate analyses of prognostic factors in patients with CCAVariableUnivariate analysisMultivariate analysisHR95% CIP valueHR95% CIP valueSex1.1870.761 to 1.8510.446 Age1.2270.802 to 1.8770.343 Tumor size (cm)1.7671.086 to 2.874 0.02* 1.5150.898 to 2.5590.120Differentiation1.6131.017 to 2.558 0.046* 1.1620.678 to 1.9920.586Perineural invasion1.6430.972 to 2.7750.054 R0 resection1.8851.129 to 3.146 0.022* 1.1620.678 to 1.9920.586Lymphatic metastasis1.8461.205 to 2.830 0.005* 1.3970.845 to 2.3110.192Vascular invasion0.9950.589 to 1.6810.984 USP1 expression3.1291.944 to 5.037 <0.001* 1.7250.925 to 3.218 <0.001* *P value<0.05.CCA, cholangiocarcinoma; USP1, ubiquitin-specific peptidase 1. Univariate and multivariate analyses of prognostic factors in patients with CCA P value<0.05. CCA, cholangiocarcinoma; USP1, ubiquitin-specific peptidase 1. To elucidate the biological role of USP1 in CCA, we performed shRNA (Sh1, Sh2, and Sh3)-mediated USP1 knockdown in RBE cells, which exhibit relatively high protein levels of USP1. Compared with those of Sh3, the transfection efficiencies of Sh1 and Sh2 were greater, indicating their suitability for subsequent experiments. Additionally, we used a plasmid to overexpress USP1 in HuCCT1 cells, which have relatively low USP1 expression (online supplemental figure S3A and B). In vitro functional analyses demonstrated that USP1 knockdown significantly suppressed the proliferation and migration capacities of CCA cells, whereas USP1 overexpression notably promoted these cellular processes (online supplemental figure S3C-F). To further confirm the regulatory effect of USP1 on fatty acid metabolism in CCA, we performed lipid metabolomics analysis in NC and USP1 knockdown RBE cells (figure 2E). A total of 104 differential lipid metabolites were identified, of which 92 were downregulated (figure 2F and G). Further MetPA enrichment analysis revealed that the downregulated lipid metabolites were enriched in glycerophospholipid metabolism, unsaturated fatty acid biosynthesis, fatty acid elongation, fatty acid degradation and fatty acid biosynthesis (figure 2H). Surprisingly, we observed that USP1 altered the levels of FFAs, especially OA, without affecting the levels of triacylglycerols (TGs) (figure 2I and J). To validate these observations, we next assessed metabolic phenotypes in vitro. The results of the FFA assay demonstrated that USP1 knockdown decreased the FFA content, while USP1 overexpression increased it (figure 2K). We subsequently performed Nile red staining and TG assays on USP1-knockdown cells and found that, compared with control cells, USP1 knockdown did not affect TG levels. Similarly, no changes in TG content were observed in USP1-overexpressing cells (figure 2L and M). FFA content in tumors is dynamically regulated by fatty acid biosynthesis, fatty acid uptake, fatty acid oxidation and lipid droplet degradation.24 25 Next, we performed oxygen consumption rate (OCR) and fatty acid oxidation (FAO) staining assays and found that USP1 expression was positively correlated with the level of FAO (figure 2N and O and online supplemental figure S3G). Further detection of ATP content yielded consistent results (figure 2P). The above experiments demonstrate that USP1 increased the FFA content and FAO level but did not affect the TG content. USP1 remodeled lipid homeostasis through maintaining the dynamic equilibrium between fatty acid synthesis and lipophagy Next, we hypothesized that the increase in the FFA content mediated by USP1 may be due to the dynamic equilibrium between lipid droplets (LDs) storage and release. Fatty acids stored in LDs are derived from endogenous synthesis and exogenous uptake.26 Fatty acid uptake assays revealed that USP1 did not affect fatty acid uptake in CCA cells (figure 3A). To further confirm the role of fatty acid biosynthesis in USP1-mediated lipid homeostasis, we transferred CCA cells to serum-free starvation (SS) medium supplemented with firsocostat (FrC). FrC is an acetyl-CoA carboxylase (ACC) inhibitor that blocks the conversion of acetyl-CoA to malonyl-CoA by inhibiting ACC activity, thereby inhibiting fatty acid biosynthesis. We subsequently assessed whether modulating USP1 altered the LDs content on inhibition of fatty acid synthesis. The results indicated that in the presence of FrC, knocking down USP1 expression significantly increased the LDs content, whereas overexpression of USP1 had the opposite effect (figure 3B and C). These findings demonstrate that impairing fatty acid synthesis disrupts the homeostasis between LDs storage and release regulated by USP1, suggesting a potential role for USP1 in promoting LD storage through the regulation of fatty acid synthesis. USP1 remodeled lipid homeostasis through maintaining the dynamic equilibrium between fatty acid synthesis and lipophagy. (A) Fatty acid uptake assays demonstrated that USP1 knockdown or overexpression did not affect fatty acid uptake capacity in CCA cells. (B) TG levels in USP1-knockdown and USP1-overexpressing cells following SS for 24 hours and FrC treatment for 6 hour. (C) Nile red staining of USP1-knockdown and USP1-overexpressing cells following SS treatment for 24 hours and FrC treatment for 6 hours. (D) ATGL and PLIN2 levels in USP1 knockdown and overexpression CCA cells treated with SS for 24 hours and with FrC for 6 hours were analyzed by Western blotting. USP1 markedly altered lipid droplet surface marker levels without affecting ATGL expression (E–F) TG levels and Nile red staining in cells with USP1 knockdown and overexpression after SS for 24 hours and after treatment with FrC and ATGLi for 6 hours. (G) USP1 knockdown and overexpression cells were cultured in SS for 24 hours and then treated with FrC and ATGLi for 6 hours, after which the cell lysates were analyzed by western blotting. USP1 maintained its regulatory capacity over CCA TG levels, indicating that USP1-mediated LDs degradation is independent of ATGL (H–I) Following culture in serum starvation medium and treatment with an FrC and ATG7i, USP1 lost its regulatory effect on CCA TG levels, demonstrating that USP1-mediated lipid droplet degradation is dependent on autophagic degradation. (J) USP1 knockdown and overexpression CCA cells were observed through transmission electron microscopy. The red arrows indicate LDs, while the black arrows indicate autolysosomes. The results showed that USP1 did not alter the number of LDs but changed the number of autolysosomes. (K) Confocal images of mRFP-GFP-LC3-expressing USP1 knockdown and overexpression CCA cells. The results showed that USP1 overexpression upregulated autophagic activity in CCA cells. (L) Western blot analyses of the expression of SQSTM1 and LC3 in USP1 knockdown and overexpression CCA cells. (M) Immunofluorescence staining demonstrated that USP1 increased LC3-lipid droplet colocalization following SS treatment for 24 hours and after FrC treatment for 6 hours. *p<0.05, **p<0.01, ***p<0.001. ATG7i, autophagy-related 7 inhibitors; ATGL, adipose triglyceride lipase; ATGLi, Atglistain; CCA, cholangiocarcinoma; FrC, firsocostat; LC3, light chain 3; LDs, lipid droplets; OA, oleate acid; PLIN2, perilipin 2; SS, serum starvation; TG, triglyceride; USP1, ubiquitin-specific peptidase 1. The release of LDs can be classified into two distinct mechanisms: lipase-mediated lipolysis and autophagy-dependent degradation. During lipolysis, adipose triglyceride lipase (ATGL) serves as the key enzyme that catalyzes the initial step of the process.27 To elucidate the specific pathway through which USP1 regulates LDs release, we assessed ATGL expression levels following USP1 knockdown or overexpression, using perilipin 2 (PLIN2) as a surface marker to monitor LDs content. PLIN2 protein expression was increased in USP1-knockdown cells but reduced in USP1-overexpressing cells. However, no alteration in the expression level of ATGL was detected (figure 3D). We subsequently treated CCA cells with ATGL and ACC inhibitors and evaluated the regulatory effect of USP1 on TG levels. The results showed that the expression level of USP1 was negatively correlated with TG content (figure 3E–G). These results indicate that USP1 promoted LD degradation independent of lipolysis. We next investigated whether USP1 regulates LD degradation through lipophagy. Autophagy-related 7 (ATG7) is a core autophagy-related protein with E1-like enzyme activity, and is responsible for facilitating the formation of autophagosomes.28 Therefore, we inhibited autophagosome formation by suppressing ATG7 and examined the regulatory role of USP1 in LDs degradation under these conditions. In CCA cells deficient in ATG7 and ACC, USP1 knockdown or overexpression resulted in a loss of the ability to regulate TG levels (figure 3H–I and online supplemental figure S4A). These findings indicated that the function of USP1 was dependent on the core autophagy machinery and fatty acid synthesis. To further validate the regulatory capacity of USP1 in lipophagy, we performed a series of autophagic characterization assays. TEM revealed that USP1 promoted the formation of autolysosomes without influencing the quantity of LDs (figure 3J). To further evaluate the impact of USP1 on autophagic flux, we transfected USP1-modulated CCA cells with a lentiviral vector expressing dual-fluorescence-tagged microtubule-associated protein 1 light chain 3 (LC3), which revealed that USP1 expression significantly increased the formation of autophagosomes and autolysosomes (figure 3K). Reduced LC3 conversion and sequestosome 1 (SQSTM1) accumulation are hallmarks of defective autophagy.29 Therefore, we measured the conversion of LC3-I to LC3-II and the expression of SQSTM1. Our analysis revealed that the knockdown of USP1 led to a reduction in LC3-II levels and an increase in SQSTM1 levels. Conversely, overexpression of USP1 had the opposite effect (figure 3L and online supplemental figure S4B). To further investigate how USP1 affected autophagic flux, we inhibited the fusion of autophagosomes and lysosomes by adding chloroquine (CQ) and subsequently evaluated the effect of USP1 on LC3-II and SQSTM1 levels (online supplemental figure S4C). The results revealed that USP1-knockdown cells showed a significantly attenuated inhibitory response to CQ on autophagy compared with the NC group. Furthermore, the combination of USP1 overexpression and CQ resulted in a greater accumulation of LC3-II than CQ alone, demonstrating its role in effectively promoting the formation of autophagosome. These findings indicate that USP1 regulated autophagy by controlling autophagosome formation. We next assessed the colocalization of LDs with autophagosomes and lysosomes to verify the occurrence of lipophagy and the regulatory role of USP1 in this process. Immunofluorescence (IF) analysis revealed that the knockdown of USP1 attenuated the binding capacity of LC3 to LDs, whereas the overexpression of USP1 potentiated this interaction (figure 3M and online supplemental figure S4D). To validate whether LDs encapsulated within autophagosomes are ultimately transported to lysosomes for degradation, we performed immunofluorescence assays examining the interaction between the LDs coating protein PLIN2 and the lysosomal marker protein LAMP1. The results demonstrated that USP1 knockdown significantly inhibited the colocalization of PLIN2 and LAMP1, whereas USP1 overexpression promoted their interaction (online supplemental figure S4E). These findings suggested that USP1 regulated fatty acid synthesis and lipophagy to maintain lipid homeostasis. Taken together, these data demonstrated that LCUFAs, which were regulated by USP1, significantly changed in CCA, but its specific effects on the TME remain unclear. The FABP5/PPARγ axis mediated TAMs uptake and metabolic utilization of CCA-derived LCUFAs To delineate immune cell populations demonstrating the most robust lipid uptake within the TME of CCA, we performed dimensionality reduction analysis on single-cell RNA sequencing data from the GSE138709 database, stratifying patient samples by histopathological tumor differentiation grade (figure 4A). Heatmap analysis of lipid metabolism-related gene sets revealed significant enrichment in TAMs, indicating their prominent lipid uptake and metabolic activity (online supplemental figure S5A). Analysis of the GSE171899 CCA single-cell transcriptome database further confirmed that compared with other immune cell subsets, TAMs exhibit significantly higher fatty acid metabolic activity (online supplemental figure S5B-D). GO enrichment analysis demonstrated a marked upregulation of lipid metabolism-related pathways in TAMs relative to normal tissue macrophages (online supplemental figure S5E). To investigate the capacity of TAMs to take up tumor-derived fatty acids, we established an in vitro three-dimensional (3D) co-culture model (figure 4B). Transcriptome sequencing was subsequently performed on the macrophages before and after co-culture. KEGG enrichment analysis revealed a total of 15 differentially upregulated pathways, among which the PPAR signaling pathway was closely associated with fatty acid metabolism (figure 4C and online supplemental figure S5F). Additionally, a marked upregulation of genes related to macrophage M2 polarization was observed (figure 4D). Previous studies have demonstrated that the ability of TAMs to sense and respond to FFAs in the TME depends on membrane transporter-mediated uptake and fatty acid-binding proteins (FABPs), which orchestrate subsequent metabolic trafficking and signaling processes.30 Further analysis of the transcriptome sequencing results revealed that co-culture induced pronounced increases in the mRNA levels of CD36 and FABP5 in TAMs (figure 4E). Single-cell sequencing analysis also revealed that FABP5 and CD36 were highly expressed in TAMs subsets. Further analysis showed that as the degree of differentiation of CCA tissue decreased, the expression level of FABP5 gradually increased, while the expression level of CD36 showed no significant change (figure 4F and online supplemental figure S5G). IF staining of clinical specimens revealed progressive correlation between FABP5 and the M2 marker cluster of differentiation 206 (CD206) during tumor progression (figure 4G and online supplemental figure S5H). Subsequently, single-cell subcluster analysis categorized TAMs into four functional subsets: FABP5+, FCN1+, CCL5+, and CD1C+ macrophages (MACs) (figure 4H). Through heatmap analysis of lipid metabolism-related genes, FABP5+ MACs were identified as a distinct TAMs subpopulation characterized by substantial lipid accumulation (online supplemental figure S5I). The proportion of these FABP5+ MACs increased progressively with tumor progression (figure 4H). These findings are consistent with the expression pattern of LCUFAs observed in our previous study across CCA tissues at different clinical stages. The above results suggest that TAMs may upregulate FABP5 expression through the uptake of CCA-derived LCUFAs. To further validate this hypothesis, we co-cultured TAMs with FrC-pretreated CCA cells and observed a significant downregulation of FABP5 expression in the TAMs (figure 4I). Furthermore, exogenous supplementation with OA and docosahexaenoic acid significantly increased FABP5 expression in TAMs (figure 4J). The FABP5/PPARγ axis mediated TAM uptake and metabolic utilization of CCA-derived LCUFAs. (A) Single-cell RNA sequencing data from patients with CCA in the GSE138709 database were analyzed by stratifying samples on the basis of tumor differentiation status. (B) Monocytes were isolated from healthy human peripheral blood and differentiated into macrophages using M-CSF, followed by co-culture with CCA cells for 48 hours to obtain TAMs. (C) Transcriptomic sequencing was performed on differentiated macrophages with or without co-culture with CCA cells for 48 hours. Volcano plot showing differentially expressed proteins (|log2FC| >1, p<0.05). KEGG enrichment analysis showing all of the significantly enriched upregulated pathways. (D) Heatmap of the PPAR pathway and markers of M2 macrophage polarization. (E) Expression of genes associated with fatty acid uptake in macrophages according to transcriptomic sequencing data. (F) Single-cell analysis of the GSE138709 dataset revealed that FABP5 was predominantly expressed on TAMs and significantly elevated with tumor progression. (G) Representative images of immunofluorescence staining for M2 macrophage markers and FABP5 expression in tumors of different grades. (H) Single-cell clustering classified the TAMs into FCN1+, FABP5+, CCL5+ and CD1C+ subsets, with the number of FABP5+ MACs proportionally increasing during tumor progression. (I) RT-qPCR confirmed that FABP5 upregulation in TAMs after co-culture was reversed by FrC treatment in CCA cells. (J) Macrophages were co-cultured with CCA cells for 48 hours to generate TAMs. Then, the TAMs were treated with 100 µmol/L OA or 50 µmol/L DHA for 24 hours. RT-qPCR showed that exogenous LCUFAs significantly increased FABP5 expression. (K) CCA cells were incubated with FITC-labeled oleic acid (FITC-OA) for 24 hours. Then, the supernatant of the CCA cells was collected to treat differentiated macrophages for 48 hours. Fluorescence microscopy was performed to observe the TAMs. (L) NC or FABP5 was knocked down in differentiated macrophages, which were then co-cultured with CCA cells for 48 hours to construct TAMs and FABP5i-TAMs. Nile red staining was performed to evaluate lipid accumulation in the TAMs. The results revealed FABP5-mediated fatty acid uptake promoted intracellular lipid accumulation in TAMs (M) Fatty acid uptake assays were performed to evaluate the fatty acid uptake capacity of TAMs. (N) A TG quantification assay was performed to evaluate the intracellular TG levels in TAMs. (O) Representative images of ROS staining in TAMs. (P–Q) Assessment of FAO in TAMs by OCR (P) and ATP (Q) measurements. The results demonstrated enhanced FAO in TAMs, reversible on FABP5 knockout. (R) Flow cytometry was performed with antibodies against CD80 and CD206 to evaluate macrophage polarization and indicated CCA-derived fatty acids drove FABP5-dependent metabolic reprogramming to promote M2 polarization in TAMs. (S) After being co-cultured with CCA cells for 48 hours, the TAMs were treated with or without T0070907 (10 µmol/L) for 24 hours. RT-qPCR was performed to measure FABP5 mRNA levels. (T) Luciferase reporter assays confirmed that FABP5 knockout markedly inhibited PPARγ transcriptional activity. *p<0.05, **p<0.01, ***p<0.001. CCA, cholangiocarcinoma; CCL, C-C motif chemokine ligand; DHA, docosahexaenoic acid; FABP5, fatty acid binding protein 5; FAO, fatty acid oxidation; FC, fold change; FrC, firsocostat; KEGG, Kyoto Encyclopedia of Genes and Genomes; LCUFAs, long-chain unsaturated fatty acids; MACs, macrophages; M-CSF, macrophage colony-stimulating factor; OA, oleate acid; OCR, oxygen consumption rate; PPAR, peroxisome proliferator-activated receptor; ROS, reactive oxygen species; RT-qPCR, Reverse Transcription Quantitative PCR; TAMs, tumor-associated macrophages; TG, triglyceride; UMAP: Uniform Manifold Approximation and Projection. Next, we investigated whether FABP5 promotes the uptake and utilization of tumor-derived LCUFAs by TAMs. Fluorescence tracing experiments using FITC-OA-preloaded CCA cells confirmed the active uptake of tumor-derived LCUFAs by TAMs, a process blocked by FABP5 knockdown in TAMs (figure 4K and online supplemental figure S5J). Consistent with these findings, Nile red staining revealed marked accumulation of LDs in co-cultured TAMs, a phenomenon similarly attenuated by FABP5 knockdown in TAMs (figure 4L). Biochemical analyses of fatty acid uptake and TG content further validated these coordinated metabolic changes (figure 4M and N). In addition, functional assays revealed that FABP5 knockdown reversed the co-culture-induced increase in ROS levels (figure 4O and online supplemental figure S5K) and reduced the OCR (figure 4P) and ATP production (figure 4Q). The above results demonstrated that TAMs uptook and used FFA derived from CCA through FABP5. We subsequently investigated the effects of FABP5-mediated fatty acid uptake on macrophage polarization, and the underlying mechanism. First, Reverse Transcription Quantitative PCR (RT-qPCR), western blotting, and dual IF co-staining for CD68/CD206 collectively demonstrated that CCA significantly increased M2 macrophage polarization, which was effectively reversed by FABP5 knockdown in TAMs (online supplemental figure S6A-C). These findings were further corroborated by flow cytometric analysis (figure 4R and online supplemental figure 6D). Furthermore, inhibition of fatty acid oxidation in TAMs with etomoxir or a reduction in intracellular ROS levels with N-acetylcysteine significantly suppressed M2 polarization of TAMs (online supplemental figure 6E and F). These results indicate that tumor-derived fatty acids upregulate FABP5 in TAMs, thereby enhancing their uptake and utilization of these fatty acids to promote M2 polarization. To investigate the specific mechanism underlying FABP5 upregulation, we treated co-cultured cells with T0070907, a PPARγ inhibitor, and observed a subsequent decrease in FABP5 expression (figure 4S). Furthermore, we found that knockdown of FABP5 similarly suppressed PPARγ transcriptional activity (figure 4T). On the basis of these findings, we concluded that TAMs use FABP5 to potentiate FAO-driven M2 polarization through a self-reinforcing positive feedback loop involving the PPARγ-mediated transcriptional activation of FABP5 itself. We further investigated the correlation between M2-polarized TAMs infiltration and clinical outcomes on TMA derived from 111 patients with CCA. Compared with those in normal tissues, the expression levels of CD206 in CCA tissues were significantly upregulated (online supplemental figure S7A). Furthermore, survival analysis and the Cox proportional hazards model demonstrated that the infiltration of M2 TAMs was significantly associated with poor prognosis in patients with CCA (online supplemental figure S7B, table 2 and online supplemental table S2). These findings collectively highlight the significant role of LCUFAs in the CCA TME. Table 2Univariate and multivariate analyses of prognostic factors in patients with CCAVariableUnivariate analysisMultivariate analysisHR95% CIP valueHR95% CIP valueSex1.1870.761 to 1.8510.446 Age1.2270.802 to 1.8770.343 Tumor size (cm)1.7671.086 to 2.874 0.02* 1.4860.886 to 2.4910.133Differentiation1.6131.017 to 2.558 0.046* 1.3550.809 to 2.2690.249Perineural invasion1.6430.972 to 2.7750.054 R0 resection1.8851.129 to 3.146 0.022* 1.4990.806 to 2.7870.201Lymphatic metastasis1.8461.205 to 2.830 0.005* 1.4460.885 to 2.3600.141Vascular invasion0.9950.589 to 1.6810.984 CD206 expression1.6421.072 to 2.515 0.024 1.5250.931 to 2.4960.094*P value<0.05.CCA, cholangiocarcinoma; CD206, cluster of differentiation 206. Univariate and multivariate analyses of prognostic factors in patients with CCA P value<0.05. CCA, cholangiocarcinoma; CD206, cluster of differentiation 206. Tumor-derived LCUFAs promoted immunosuppression in CCA by upregulating FABP5+ Lipid-loaded TAMs To investigate whether USP1 remodeled the immunosuppressive microenvironment through fatty acid metabolic reprogramming, we analyzed data from the TIMER database (http://timer.cistrome.org/). The results confirmed a positive correlation between USP1 expression and TAM M2 polarization (online supplemental figure S8A). The results of the TMA analysis revealed a significant positive correlation between USP1 and CD206 expression levels in CCA tissues (online supplemental figure S8B). Survival analysis demonstrated that patients with high expression of both USP1 and CD206 had a significantly worse prognosis than those with concurrent low expression or high expression of either marker alone did, suggesting the potential prognostic value of this biomarker combination (online supplemental figure S8C). Thereafter, we generated hepatocyte-specific USP1 knockout mice via Adeno-associated virus (AAV)-mediated gene deletion and established spontaneous CCA models using hydrodynamic injection (figure 5A). Successful model construction was confirmed by immunohistochemical staining (figure 5B). Analysis of tumor progression demonstrated that USP1 knockout significantly suppressed tumor burden, reduced tumor size, and extended survival duration (figure 5C–E). We then hypothesized that USP1 regulates lipid homeostasis to modulate TAMs function. IF staining and flow cytometry further verified that USP1 knockdown markedly inhibited the M2 polarization of TAMs (figure 5F and G and online supplemental figure S9A and B). Tumor-derived LCUFAs promoted immunosuppression in CCA by upregulating FABP5+ lipid-loaded TAMs. (A) USP1 gene-edited mice were generated via AAV, and a spontaneous CCA model was established using hydrodynamic-based tumor induction. (B) H&E, CK19, and USP1 staining of tumor tissues confirmed successful model construction. (C–D) Mice were sacrificed at 5 weeks after successful hydrodynamic injection, and tumors were harvested for comparison of number (C) and size (D). (E) Overall survival of mice with spontaneously formed CCA. (F) A single-cell suspension was prepared from mouse tumor tissues, followed by flow cytometry analysis with corresponding antibodies. The results confirmed USP1-mediated regulation of TAMs M2 polarization. (G) Representative multiplexed IF staining images. (H) Single-cell RNA sequencing was performed on tumor tissues from NC and sh-USP1 spontaneous tumor-bearing mice. (I) USP1 knockout significantly suppressed FABP5 expression in TAMs without altering CD36 or FABP4 levels. (J) Dimensionality reduction clustering classified TAMs into FABP5+ MACs, FOLR2+ MACs, LY6C2+ MACs, MACs C4, and MACs C5 subsets. USP1 knockdown markedly reduced the proportion of FABP5+ MACs. (K) Pathway enrichment analysis revealed that compared with FABP5− MACs, FABP5+ MACs exhibited enhanced lipid metabolism and promoted an immunosuppressive microenvironment. (L–M) Dimensionality reduction of T cells and intercellular communication analysis revealed that compared with FABP5− MACs, FABP5+ MACs had significantly weaker interactions with cytotoxic T cells. (N) Pathway analysis indicated that compared with FABP5− MACs, FABP5+ MACs exhibited attenuated interactions with effector T cells via the CCL4 and MHC-I pathways. (O) USP1 knockout further suppressed TAMs-cytotoxic T-cell interactions, with a more pronounced inhibitory effect on FABP5+ TAMs. (P) Macrophages were co-cultured with either control CCA cells or USP1-knockdown CCA cells for 48 hours. After the culture medium was replaced, the conditioned medium from the TAMs was collected and subsequently co-cultured with activated CD8+ T cells isolated from peripheral blood. T-cell proliferation was then assessed using CFSE staining. (Q) The activity of CD8+ T cells was validated by measuring IFN-γ production. *p<0.05, **p<0.01, ***p<0.001. CCA, cholangiocarcinoma; CCL, C-C motif chemokine ligand; CFSE, Cell Proliferation Tracer Fluorescent Probe; FABP5, fatty acid binding protein 5; IF, immunofluorescence; IFN, interferon; LCUFAs, long-chain unsaturated fatty acids; MACs, macrophages; MHC, major histocompatibility complex; TAMs, tumor-associated macrophages; USP1, ubiquitin-specific peptidase 1; UMAP, Uniform Manifold Approximation and Projection. To determine the impact of USP1 on the immune microenvironment and its potential dependence on LCUFAs, we performed single-cell RNA sequencing on tumor tissues from both the NC and the Sh-USP1 groups. The results revealed significant enrichment of lipid metabolism-related genes in TAMs (figure 5H and online supplemental figure S9C). FABP5 expression was significantly upregulated in TAMs compared with that in other cell types, which was consistent with previous single-cell data from patients with CCA (online supplemental figureS 9D). Notably, USP1 knockout specifically suppressed FABP5 expression in TAMs without affecting CD36 or FABP4 levels (figure 5I and online supplemental figure S9E). Dimensionality reduction analysis revealed five TAMs subsets, including FABP5+, LY6C2+, and FOLR2+ clusters (figure 5J and online supplemental figure S9F). GO enrichment analysis demonstrated that FABP5+ MACs exhibited activated lipid metabolism and immunosuppressive pathways alongside suppressed immune activation signals, with USP1 knockdown significantly reducing the proportion of this subset (figure 5K). These results suggest that USP1 may increase lipid accumulation in TAMs, drive M2 polarization, and reshape the immunosuppressive TME by regulating FABP5 expression in TAMs. On the basis of these findings, we established an in vitro co-culture system to further validate this mechanism. USP1 knockdown suppressed FABP5 expression, carnitine palmitoyltransferase 1A levels, and PPAR pathway activity in TAMs, and these effects were reversed by exogenous OA supplementation (online supplemental figure S10A). USP1 knockdown in CCA cells significantly suppressed fatty acid uptake and LD accumulation in TAMs, whereas FABP5 overexpression in TAMs completely reversed these effects (online supplemental S10B-D). OCR and ATP production assays revealed that USP1 knockdown markedly inhibited FAO activity in TAMs, and that FABP5 overexpression restored oxidative metabolic activity (online supplemental figure S10E and F). ROS level measurements further corroborated the consistency of this regulatory mechanism (online supplemental figure S10G). IF staining demonstrated that USP1 knockdown significantly inhibited the M2 polarization of TAMs, whereas specific overexpression of FABP5 in TAMs effectively reversed this inhibitory effect (online supplemental figure S10H). Flow cytometry analysis further confirmed these results (online supplemental figure S10I). These findings confirmed the regulatory role of USP1 in FABP5+ MACs. To further investigate the impact of USP1-regulated FABP5+ MACs on T-cell function, we performed dimensionality reduction analysis on T cells and identified 7 T-NK subsets, including cytotoxic T cells, memory T cells, natural killer (NK) cells, proliferating T cells, effector T cells, helper T cells and regulatory T cells (figure 5L). Compared with their FABP5− counterparts, FABP5+ MACs displayed significantly weakened interactions with cytotoxic T cells, as characterized by the inhibition of CCL4 signaling and major histocompatibility complex (MHC-I) signaling (figure 5M and N). USP1 weakened this communication impairment and substantially increased cytotoxic T-cell infiltration and activation by suppressing the proportion and functional output of FABP5+MACs (figure 5O). To further confirm the regulatory role of USP1 on cytotoxic T cells via TAMs, we subsequently co-cultured CCA cells (NC group vs Sh-USP1 group) with TAMs for 48 hours. The conditioned medium from the TAMs was collected and subsequently co-cultured with CD8+ T cells to validate the regulatory effect of USP1-modulated TAMs on CD8+ T-cell function. Cell Proliferation Tracer Fluorescent Probe (CFSE) proliferation assays revealed that TAMs significantly suppressed the proliferation of CD8+ T cells. Notably, USP1 knockdown in CCA cells inhibited fatty acid uptake by TAMs through the downregulation of FABP5 expression, thereby enhancing the proliferation of CD8+ T cells (figure 5P and online supplemental figure S11A). Detection of IFN-γ also revealed the regulatory role of USP1 in CD8+ T cells activation (figure 5Q). Collectively, these findings suggest that USP1 can mediate the release of CCA-derived LCUFAs to increase the proportion of FABP5+ MACs, thereby suppressing cytotoxic T-cell function through fatty acid metabolic reprogramming in TAMs. USP1 interacted with ACLY and ATG3 To elucidate the potential mechanisms of USP1-mediated fatty acid synthesis and lipophagy, we performed immunoprecipitation (IP)-MS analysis to explore the proteins that interact with USP1 and bind to genes related to this pathway. The results revealed ATG3 and ACLY as potential targets that interact with USP1 (figure 6A and B). ACLY plays a crucial role in regulating fatty acid synthesis by providing the essential precursor acetyl-CoA.31 32 ATG3 is a critical component of the autophagy pathway, ensuring the proper formation and function of autophagosomes.33 34 Thus, we speculated that ATG3 and ACLY might be required for the USP1-mediated regulation of lipid homeostasis. IF staining of RBE cells and CCA tissues revealed that USP1 was spatially colocalized with both ATG3 and ACLY (figure 6C and D). Endogenous ATG3 and ACLY were detected in USP1 immunoprecipitates from RBE cells (figure 6E). Subsequently, GST pull-down assays validated the direct interaction of USP1 with ATG3 and ACLY (figure 6F). Furthermore, we simulated the 3D structure of the USP1/ACLY and USP1/ATG3 complex. The docking results revealed the possible amino acids in the region of USP1 that interact with the possible amino acids in the regions of ATG3 and ACLY (figure 6G). To investigate the specific region responsible for the interaction between USP1, ATG3 and ACLY, we generated truncated mutant fragments of these three proteins (figure 6H). Our transfection experiments revealed that the deletion of amino acids 401–785 from USP1 abolished its ability to bind to ACLY and ATG3, whereas the deletion of amino acids 1–200 or 1–400 had no effect in HEK 293T cells (figure 6I). In addition, deletion of amino acids 1–620 from ACLY and amino acids 1–121 from ATG3 abolished the ability of the proteins to bind to USP1 in HEK 293T cells (figure 6J and K). These experiments delineate the critical reciprocal binding regions of each protein. USP1 interacted with ACLY and ATG3. (A–B) IP/MS screening of USP1-interacting proteins revealed ATG3 and ACLY as binding partners. (C–D) Immunofluorescence staining of RBE cells and tissues confirmed the spatial colocalization of USP1 and ATG3/ACLY. (E) Co-IP assays were performed to explore the interaction between USP1 and ATG3/ACLY in RBE cells. (F) GST pull-down assays were performed to examine the direct interaction between USP1 and ATG3/ACLY. (G) Molecular docking predicted potential binding interfaces among these proteins. (H) Design of USP1, ACLY and ATG3 truncations. (I) IP assays following the transfection of USP1 truncation mutants in HEK 293T cells revealed that the 201–400 amino acid residues of USP1 mediate its interaction with ACLY and ATG3. (J) Transfection of truncation mutants of ACLY in HEK 293T cells demonstrated that its N-terminal domain (residues 1–620) is essential for USP1 binding. (K) Similarly, transfection of ATG3 truncation mutants identified its N-terminal region (residues 1–120) as the critical binding domain for USP1. *p<0.05, **p<0.01, ***p<0.001. ACLY, glutathione S-transferase; ATG3, autophagy related 3; Co-IP, co-immunoprecipitation; GST, glutathione S-transferase; IP/MS, immunoprecipitation/mass spectrometry; USP1, ubiquitin-specific peptidase 1. USP1 facilitated K48-Linked deubiquitination of ACLY and ATG3 To validate the hypothesis that USP1 directly stabilized ATG3 and ACLY by preventing their degradation, we knocked down and overexpressed USP1 in CCA cells. The results revealed that USP1 positively regulated the protein levels of ACLY and ATG3 (figure 7A and B). Considering that USP1 is a deubiquitinase that can interact with ATG3 and ACLY, we then investigated whether USP1 might stabilize ATG3 and ACLY. Overexpression of WT USP1 increased the level of ATG3 and ACLY, whereas overexpression of the catalytically inactive C90A mutant USP1 did not cause such an increase (figure 7C). Additionally, the proteasome inhibitor MG132 reversed the USP1 knockdown-induced decrease in ATG3 and ACLY levels (figure 7D). Moreover, with the protein synthesis inhibitor cycloheximide (CHX), USP1 maintained the stability of ATG3 and ACLY (figure 7E and F). These experiments suggested that USP1 regulated ATG3 and ACLY stability is dependent on the ubiquitin proteasome system. To directly evaluate the impact of USP1 on the enzymatic activity of its substrates, we measured the activity of ACLY in USP1-knockdown RBE cells. USP1 knockdown significantly inhibited ACLY activity (online supplemental figure S11B). Furthermore, after autophagy was induced by EBSS starvation, LC3 lipidation was markedly impaired in USP1-knockdown cells (online supplemental figure S11C). These findings indicate that USP1 regulates the functional enzymatic outputs by decreasing the protein levels of ACLY and ATG3. USP1 facilitated K48-linked deubiquitination of ACLY and ATG3. (A–B) The protein expression levels of ATG3 and ACLY were detected in USP1-knockdown RBE cells (A) and USP1-overexpressing HuCCT1 cells (B). The results showed USP1 positively regulated ATG3 and ACLY expression. (C) Increasing amounts of His-tagged USP1 (WT or C90A mutant) were transfected into HEK 293T cells. Western blot assays were performed to analyze the expression of ATG3 and ACLY and revealed that WT-USP1 upregulated ATG3 and ACLY expression in a dose-dependent manner, whereas the C90A mutant completely lost this regulatory capability. (D) Western blotting of ATG3 and ACLY expression in cells treated with MG132 (10 µmol/L) for 6 hours. The results showed that USP1 knockdown significantly inhibited ATG3 and ACLY protein expression, and this inhibition could be reversed by treatment with the proteasome inhibitor MG132. (E–F) ACLY and ATG3 levels in transfected RBE and HuCCT1 cells treated with CHX (50 µg/mL) for the indicated times. The protein levels of ACLY and ATG3 were quantified. The results demonstrated that USP1 knockdown significantly shortened the protein half-life of ATG3 and ACLY, while USP1 overexpression extended their stability. (G–H) Ubiquitination levels of ATG3 (G) and ACLY (H) in transfected CCA cells treated with 10 µM MG132 for 6 hours. USP1 knockdown markedly enhanced polyubiquitination levels of ATG3 and ACLY, whereas USP1 overexpression suppressed their ubiquitination. (I–J) HA-tagged Ub and His-tagged USP1 (WT or C90A mutant) were transfected into HEK 293T cells, after which the ubiquitination levels of ATG3 (I) and ACLY (J) were detected after treatment with MG132 (10 µM) for 6 hours. The study revealed that transfection with the USP1-C90A mutant failed to regulate the ubiquitination modifications of ATG3 and ACLY as effectively as wild-type USP1. (K–L) HEK 293T cells were cotransfected with Myc-ATG3 or Flag-ACLY, together with His-USP1 and HA-tagged Ub (Lys0, Lys48 alone, or Lys63 alone). Following treatment with 10 µmol/L MG132 for 6 hours, the ubiquitination of ATG3 (K) and ACLY (L) was assessed. USP1 specifically regulates the ubiquitination-mediated degradation process of ATG3 and ACLY through K48-linked ubiquitin chains. *p<0.05, **p<0.01, ***p<0.001. ACLY, glutathione S-transferase; ATG3, autophagy related 3; CCA, cholangiocarcinoma; CHX, cycloheximide; USP1, ubiquitin-specific peptidase 1. Next, we transfected HA-tagged ubiquitin into CCA cells and then administered MG132 to inhibit ATG3 and ATG3 degradation. We observed increased ubiquitination levels of ACLY and ATG3 on knockdown of USP1, whereas overexpression of USP1 decreased their ubiquitination levels (figure 7G and H). Consistent with the results of the above experiments, compared with USP1 overexpression, the inactive C90A mutant lost the ability to regulate the deubiquitination of ACLY and ATG3 in HEK 293T cells (figure 7I and J). We next explored the effect of USP1 on the polyubiquitin modification of the ATG3 and ACLY proteins. The results revealed that USP1 cleaves the polyubiquitin chain associated with Lys48 but has no significant effect on the polyubiquitin chain associated with Lys63 in HEK 293T cells (figure 7K and L). In summary, these results demonstrated that USP1 stabilized the protein expression of ATG3 and ACLY through K48-linked ubiquitination chain-mediated deubiquitination activity. Deubiquitinases function by modifying specific lysine residues on their substrate proteins. Through analysis using the GPS-Uber (http://gpsuber.biocuckoo.cn/) and CPLM (https://cplm.biocuckoo.cn/) databases, we predicted five common potential ubiquitination sites on the ATG3 protein. K9O, K11O, K710, K131O, and K185O indicate the ATG3 protein with intact Lys9, Lys11, Lys71, Lys131, and Lys185 residues alone, respectively. Western blotting revealed that K9 was the key site on ATG3 that were deubiquitinated by USP1 (online supplemental figure S11D). However, further investigations are needed to determine whether USP1 promotes CCA progression by coordinating the ATG3-dependent autophagy pathway and ACLY-mediated lipid metabolic reprogramming to remodel tumor lipid homeostasis. USP1 modulated tumor progression and lipid homeostasis via regulation of ATG3 and ACLY To investigate whether ATG3 and ACLY function as key downstream effectors of USP1, we performed USP1 knockdown combined with ACLY overexpression, ATG3 overexpression, or concurrent ACLY and ATG3 overexpression in CCA cells. We observed that the overexpression of ATG3 significantly counteracted the inhibitory effects of USP1 knockdown on autophagy (figure 8A and B and online supplemental figure S11F). Nile red staining demonstrated that ATG3 overexpression enhanced lipophagy, whereas ACLY overexpression increased fatty acid biosynthesis in USP1-knockdown CCA cells (figure 8C). When ATG3 and ACLY were overexpressed simultaneously, no significant difference in TG levels was observed in USP1-knockdown CCA cells (figure 8C). FAO staining and OCR assays revealed that the reduction in FAO levels caused by USP1 knockdown can be partially reversed by the overexpression of ACLY or ACLY alone. However, when ATG3 and ACLY were overexpressed simultaneously, the FAO levels were fully restored (figure 8D and E). Further detection of ATP content yielded consistent results (figure 8F). USP1 modulated tumor progression and lipid homeostasis via regulation of ATG3 and ACLY. (A) Confocal images of mRFP-GFP-LC3-expressing RBE cells following USP1 knockdown in the presence or absence of ATG3 overexpression. (B) Western blot analysis detected the expression of SQSTM1 and LC3 in RBE cells following transfection with USP1 knockdown in the presence or absence of ATG3 overexpression. The results showed the inhibitory effect of USP1 knockdown on autophagic flux could be reversed by ATG3 overexpression. (C–F) Stable RBE cell lines were established with USP1 knockdown alone or in combination with ACLY and/or ATG3 overexpression. Nile red staining (C), FAOBlue staining (D), OCR measurement (E), and ATP content assays (F) of established RBE cells were subsequently performed. (G–I) RBE-related stable cells (control, Sh-USP1, Sh-USP1+ATG3, Sh-USP1+ACLY and Sh-USP1+ACLY+ATG3) were subcutaneously injected into the groin region of nude mice. Images (G), weights (H) and volumes (I) of subcutaneous tumors. (J) Representative images of H&E staining, IHC staining for USP1 and Nile red staining in xenograft tissues. (K) Representative images of mice injected intravenously with stably transfected RBE cells. Tumor growth was evaluated via bioluminescence 4 weeks after injection. (L) Relative fluorescence intensity of lung metastatic tumors established with stably transfected RBE cells. The results demonstrated that USP1 governs tumor invasion and metastasis via ATG3 and ACLY. (M) Representative images of orthotopic tumors established with stably transfected RBE cells. (N) Relative maximum tumor size of orthotopic tumors. The results revealed USP1 knockout significantly inhibited in situ tumor growth, which could be reversed by ATG3 and ACLY overexpression. *p<0.05, **p<0.01, ***p<0.001. ACLY, glutathione S-transferase; ATG3, autophagy related 3; FAO, fatty acid oxidation; OCR, oxygen consumption rate; USP1, ubiquitin-specific peptidase 1. We subsequently examined the effects of ATG3 and ACLY on the malignancy of CCA driven by USP1. In vitro experiments demonstrated that USP1 knockdown significantly inhibited the proliferation and migration abilities of CCA cells. The overexpression of ATG3 or ACLY alone partially reversed the inhibitory effects mediated by USP1 knockdown, whereas the co-overexpression of both genes completely reversed the cellular phenotypes (online supplemental figure S12). In vivo experiments showed consistent results. Subcutaneous xenograft models revealed that USP1 knockdown markedly suppressed tumor growth, and this suppression was partially counteracted by ACLY or ATG3 overexpression alone and completely abolished by the co-overexpression of both genes (figure 8G–I). IHC analysis confirmed that USP1 knockdown reduced the protein expression levels of Ki67 and PCNA, whereas Nile red staining revealed decreased LD deposition (figure 8J and online supplemental figure S13). These changes tended to reverse with increasing ACLY/ATG3 expression levels. Additionally, lung metastasis models further demonstrated that USP1 knockdown inhibited distant metastasis, whereas ACLY/ATG3 overexpression restored metastatic capabilities (figure 8K–L). Orthotopic liver transplantation tumor models revealed that USP1 regulated tumor growth at the primary site via ATG3 and ACLY (figure 8M–N). These results demonstrate that USP1 drove the malignant progression of CCA through synergistic regulation of ATG3-mediated lipophagy and ACLY-dependent lipogenesis. Overcoming immunotherapy resistance in CCA through targeting USP1, PD-L1 and TAMs In this study, we observed that co-culture with TAMs significantly upregulated USP1 mRNA expression levels in CCA cells (figure 9A). The TIMER database (http://timer.cistrome.org/) also revealed a positive correlation between USP1 expression and TAMs infiltration (online supplemental figure S14A). TMA revealed that high TAMs infiltration indicated poor prognosis (online supplemental figure S14B) and table 3). Further analysis revealed a positive correlation between CD68 and USP1 expression (online supplemental figure S14C and D). Survival analysis revealed that patients with high USP1 and CD68 expression had a significantly poorer prognosis (online supplemental figure S14E and online supplemental table S3). On the basis of these findings, we investigated whether TAMs promote the progression of CCA through the upregulation of USP1 expression. Colony formation and Transwell migration assays confirmed that TAMs enhanced CCA cell proliferation and migration via USP1 induction (online supplemental figure S15). Overcoming immunotherapy resistance in CCA through targeting of USP1, PD-L1 and TAMs. (A) RT-qPCR analysis showed that after co-culture with TAMs for 48 hours, USP1 mRNA expression was significantly upregulated in CCA cells. (B) Onlink Target 96 analysis was performed in RBE cells with or without co-culture with TAMs for 48 hours. Volcano plot showing differentially expressed proteins (p<0.05). (C) KEGG enrichment analysis revealed the top 20 most significantly enriched pathways. (D) After exogenous addition of recombinant CCL3 (160 ng/mL) and MMP1 (100 ng/mL) proteins to CCA cells, RT-qPCR was performed to detect USP1 mRNA expression. The results showed CCL3 specifically activated USP1 expression in CCA cells, while MMP1 showed no regulatory effect. (E) ELISA analysis of CCL3 levels in supernatants from monocultured CCA cells and the supernatants from CCA cells or TAMs isolated from the co-culture system. (F) CCA cells were co-cultured with or without CCL3-knockdown TAMs, and the intracellular CCL3 levels in the CCA cells were measured using an ELISA kit. The results confirmed that CCL3 upregulation originated from TAMs paracrine signaling. (G) After exogenous recombinant CCL3 (160 ng/mL) was added, CCA cells were treated with either Maraviroc (1 µmol/L) or BX471 (1 µmol/L) for 24 hours. USP1 mRNA levels were detected by RT-qPCR. Maraviroc, rather than BX471, significantly decreased USP1 mRNA level. (H) Representative images of dual IF staining in CCA cells with or without TAM co-culture. (I) Representative images of IF staining of clinical CCA tissue specimens. (J) Changes in USP1 and β-catenin expression in RBE cells in response to CCL3 knockdown in TAMs or treatment of RBE cells with 1 µM maraviroc for 24 hours were detected by western blotting. (K) A humanized NOG mouse model was successfully established, and flow cytometry confirming the effective engraftment of human immune cells in peripheral blood. Subcutaneous xenograft models were generated to evaluate the efficacy of combination therapy with IgG (I) or an anti-PD-L1 antibody (P), a USP1 inhibitor (U), and a macrophage-depleting agent (C). (L–N) Tumor images (L), weights (M) and volumes (N) of subcutaneous tumors in humanized NOG mice. (O) Mouse body weights were not significantly affected by the different drug treatments. (P) Flow cytometry was used to evaluate the proportions of infiltrating TAMs and M2-TAMs in the combination treatment groups. (Q) Representative images of multiplex immunohistochemical staining. *p<0.05, **p<0.01, ***p<0.001. CCA, cholangiocarcinoma; CCL3, C-C motif chemokine ligand 3; IF, immunofluorescence; KEGG, Kyoto Encyclopedia of Genes and Genomes; MMP1, matrix metallopeptidase 1; mRNA, messenger RNA; PBMCs, peripheral blood mononuclear cells; PBS, phosphate-buffered saline; PD-L1, programmed cell death-Ligand 1; RT-qPCR, Reverse Transcription Quantitative PCR; TAMs, tumor-associated macrophages; USP1, ubiquitin-specific peptidase 1. Table 3Univariate and multivariate analyses of prognostic factors in patients with CCAVariableUnivariate analysisMultivariate analysisHR95% CIP valueHR95% CIP valueSex1.1870.761 to 1.8510.446 Age1.2270.802 to 1.8770.343 Tumor size (cm)1.7671.086 to 2.874 0.02* 1.4260.853 to 2.3840.176Differentiation1.6131.017 to 2.558 0.046* 1.3450.809 to 2.2360.253Perineural invasion1.6430.972 to 2.7750.054 R0 resection1.8851.129 to 3.146 0.022* 1.1620.850 to 2.9010.149Lymphatic metastasis1.8461.205 to 2.830 0.005* 1.5330.942 to 2.4970.086Vascular invasion0.9950.589 to 1.6810.984 CD68 expression1.9551.275 to 2.996 0.002* 1.4770.894 to 2.4400.128*P value<0.05.CCA, cholangiocarcinoma; CD68, cluster of differentiation 68. Univariate and multivariate analyses of prognostic factors in patients with CCA P value<0.05. CCA, cholangiocarcinoma; CD68, cluster of differentiation 68. To further elucidate the underlying molecular mechanism, we performed an Onlink Target 96 array (online supplemental figure S16A). The results revealed marked upregulation of matrix metallopeptidase 1 (MMP-1) and CCL3 expression in co-cultured CCA cells (figure 9B and online supplemental figure S16B). KEGG pathway analysis further revealed the activation of chemokine signaling and cytokine–cytokine receptor pathways (figure 9C). Notably, exogenous supplementation experiments demonstrated that recombinant CCL3, but not MMP1, specifically induced USP1 mRNA upregulation (figure 9D). To determine the origin of CCL3, ELISAs revealed a gradient increase in CCL3 secretion levels across distinct culture conditions: supernatants from CCA cells alone, co-cultured CCA cells, and co-cultured TAMs (figure 9E). The results showed that CCL3 might mainly derive from TAMs. Subsequently, knockdown of CCL3 in TAMs significantly reduced intracellular CCL3 levels in co-cultured CCA cells, indicating that TAM-derived CCL3 is transferred to tumor cells via paracrine signaling (figure 9F). Given that CCL3 belongs to the CC chemokine family, we investigated its canonical receptors C-C motif chemokine receptor 1/C-C motif chemokine receptor 5 (CCR1/CCR5). Receptor antagonist assays revealed that CCR5 inhibition, but not CCR1 blockade, markedly reversed the CCL3-induced upregulation of USP1 expression (figure 9G). IF colocalization and clinical tissue analysis confirmed enhanced membrane colocalization of CCL3 and CCR5 in co-cultured CCA cells, which was recapitulated in clinical specimens (figure 9H and I). In glioblastoma, beta-catenin (β-catenin) was reported to activate USP1 transcription.35 Chen et al demonstrated that in triple-negative breast cancer, CCR5 upregulates β-catenin expression via YAP1.36 On the basis of these findings, we hypothesized that TAMs may upregulate USP1 expression in CCA cells through the CCL3/CCR5/β-catenin signaling axis. Western blot analysis revealed that TAMs-derived CCL3 significantly upregulated β-catenin and USP1 protein levels through CCR5 activation, whereas both CCL3 knockdown in TAMs and CCR5 inhibition in CCA cells abolished these effects (figure 9J). Further Chromatin Immunoprecipitation-Quantitative PCR (ChIP-qPCR) assays revealed that β-catenin specifically binds to the promoter region of USP1 (online supplemental figure S16C). Knockdown of β-catenin expression significantly suppressed the mRNA and protein expression of USP1 (online supplemental figure S16D and E). To evaluate the immunomodulatory role of USP1, we established a humanized NOG mouse model, which was validated by flow cytometry, confirming the successful engraftment of human CD45+ cells (figure 9K). The use of macrophage depletion combined with PD-L1 inhibition significantly suppressed subcutaneous tumor growth, with enhanced therapeutic efficacy on USP1 inhibitor addition (figure 9L–N). Furthermore, no significant changes in body weight were observed across all groups, indicating minimal severe adverse effects from these regimens (figure 9O). Flow cytometry analysis of tumor tissues revealed that macrophage depletion combined with a PD-L1 inhibitor effectively reduced TAMs infiltration. The addition of a USP1 inhibitor further reversed TAMs M2 polarization, thereby activating an antitumor immune microenvironment (figure 9P and Supplementary S16F). Furthermore, multiplex fluorescence immunohistochemical analysis revealed that triple therapy with a USP1 inhibitor, macrophage depletion and a PD-L1 inhibitor suppressed TAM lipid uptake, thereby reducing TAM infiltration and M2 polarization while promoting CD8+T cell recruitment (figure 9Q and online supplemental figure S16G). Discussion Through multi-omics sequencing, this study revealed abnormal lipid metabolism in CCA tissues, demonstrating that the accumulation of LCUFAs, represented by OA, is significantly correlated with lymphatic metastasis, liver invasion and tumor size. This finding aligns with reported protumorigenic roles of LCUFAs in various solid tumors,37 38 suggesting its potential as a key metabolic mechanism driving CCA malignancy. Therefore, the core molecular mechanisms underlying fatty acid metabolic reprogramming in the CCA TME were subsequently explored. Mechanistic investigations revealed that the deubiquitinating enzyme USP1, which is highly expressed in CCA, stabilizes ACLY and ATG3 in an UPS-dependent manner, coordinately regulating the dynamic balance between lipid synthesis and lipophagy to maintain metabolic homeostasis of LCUFAs in CCA cells. Previous studies have demonstrated that tumor-derived fatty acids can be released into the TME and subsequently taken up by various immune cells.39 40 Our study revealed that TAM exhibited the highest fatty acid uptake capacity within the CCA TME. Further analysis demonstrated that LCUFAs released by CCA cells drive lipid uptake and oxidative metabolism in TAMs through the FABP5/PPARγ signaling axis, promoting M2 polarization and shaping an immunosuppressive microenvironment. Single-cell sequencing and a spontaneous tumor model in mice confirmed that USP1 deletion significantly reduced the proportion of FABP5+ MAC subsets and restored cytotoxic T-cell function through MHC-I and CCL4 pathway. Within the TME, the synergistic action of MHC-I-mediated cytotoxic T cells activation41 42 and CCL4-directed recruitment43 is essential for effective antitumor immunity. Disruption of both pathways markedly impairs cytotoxic T-cell proliferation and activation. On the basis of these findings, we conducted preclinical experiments demonstrating that targeting USP1 combined with anti-PD-L1 therapy and TAMs depletion inhibited CCA progression by coordinately regulating fatty acid metabolic reprogramming and the immune microenvironment. This work elucidates a mechanism driving CCA malignancy through the “metabolic–immune” axis, providing a theoretical foundation for combined metabolic intervention and immunotherapy. Metabolic reprogramming plays a critical role in sustaining tumor cell proliferation and microenvironmental adaptation.44 45 In this study, we found markedly reduced LCUFAs levels in early-stage tumors and compensatory increases in advanced-stage tumors. This phenomenon likely involves multistage metabolic reprogramming mechanisms. The early depletion of LCUFAs may result from their extensive mobilization for membrane phospholipid biosynthesis during rapid tumor proliferation.24 Moreover, the activated β-oxidation pathway critically fuels tumor cell proliferation by catabolizing substantial amounts of LCUFAs to meet the bioenergetic demands of rapid growth.46 During tumor progression, increased intratumoral hypoxia activates HIF-1α to coordinate metabolic adaptation to promote malignant progression.47 The HIF-1α signaling pathway drives intracellular FFA accumulation in cancer cells by upregulating fatty acid uptake or de novo fatty acid synthesis,48 which was also demonstrated in our previous study of CCA.19 The precise mechanisms require further validation via spatiotemporal omics approaches. Although USP1 knockdown inhibited FFA release from the CCA, it did not significantly alter CD36 expression in TAMs. Notably, FABP5 protein levels were markedly downregulated, suggesting the involvement of distinct regulatory mechanisms within the fatty acid metabolic network. As a transmembrane transporter, CD36 expression may be stably regulated by transcription factors, noncoding RNAs, and post-translational modifications of proteases.49 However, FFAs regulate CD36 mainly through triggering its internalization.50 In contrast, the expression and activity of FABP5 are dynamically modulated by the intracellular fatty acid metabolic status. In hepatocellular carcinoma, a previous study demonstrated that LCUFAs released by tumor cells activate PPARγ via FABP5 in TAMs.51 In this study, we further demonstrated that TAMs uptake LCUFAs derived from CCA cells through a PPARγ-FABP5 positive feedback regulatory loop. The resistance of CCA to immunotherapy is correlated with its complex immunosuppressive microenvironment: histological features include significant CD8+ T-cell infiltration deficiency,52 M2-polarized TAM infiltration53 54 and hyperactivation of regulatory T cells.52 These multidimensional immunosuppressive characteristics collectively shape the typical “cold tumor” phenotype of CCA, limiting the efficacy of traditional immune checkpoint inhibitors. In this study, we observed that the mechanisms underlying TAMs-mediated remodeling of the immunosuppressive microenvironment involve multidimensional features. First, LCUFAs promote M2 polarization of TAMs by globally increasing cellular fatty acid uptake, β-oxidation capacity, and ROS levels. Second, LCUFAs specifically induce the infiltration and expansion of FABP5+ lipid-loaded MACs, which downregulate MHC-1 and CCL4 signaling to shape an immune-evasion TME. Accordingly, we propose a triple-combination therapy targeting USP1 inhibitors, clodronate liposomes, and anti-PD-L1, which demonstrated combined antitumor effects in preclinical models. First, USP1 inhibitors modulate the tumor metabolic microenvironment by promoting the ubiquitination degradation of fatty acid synthases while inhibiting lipophagy-related protein stabilization, thus significantly reducing LCUFA accumulation. Notably, although LCUFAs do not alter overall TAM infiltration, they induce FABP5+ MAC subpopulation expansion and M2 polarization, subsequently suppressing cytotoxic T-cell proliferation and activation. Second, although clodronate liposomes effectively deplete TAM subsets, they cannot completely eliminate all TAMs. In this context, the addition of USP1 knockout further inhibits the function of residual TAMs. Finally, anti-PD-L1 therapy significantly enhances cytotoxic T-cell cytotoxicity after the immunosuppressive barrier is dismantled, resulting in the formation of a cascade of “metabolic regulation–immune remodeling–checkpoint unlocking”. Collectively, our data demonstrate that USP1 knockdown suppresses tumor progression through dual mechanisms: a TAM-independent metabolic pathway (inhibiting fatty acid oxidation) and a TAM-dependent immune pathway (inhibiting TAM polarization). This strategy provides a novel therapeutic paradigm to overcome immunotherapy resistance in CCA. However, this study has several limitations that should be addressed in future research, including the lack of absolute quantification of LCUFAs using stable isotope-labeled standards and insufficient evidence for the specificity of lipophagy. Meanwhile, this study did not directly verify the impact of ATG3 and ACLY overexpression on the TAM phenotype in vivo. Future studies should further evaluate their potential effects on the TME in immunocompetent models. Conclusions This study revealed that USP1 orchestrates lipid metabolic reprogramming in CCA by driving LCUFA accumulation through dual pathways, which fuels FABP5+ MAC-mediated immunosuppression via M2 polarization and suppression of cytotoxic T-cell activity. Mechanistically, a self-reinforcing TAM-tumor feedback loop dependent on CCL3/CCR5/β-catenin signaling perpetuates this immunosuppressive microenvironment. Notably, a triple-combination strategy targeting USP1, depleting TAMs and blocking PD-L1 demonstrated potent antitumor effects (figure 10). These findings reveal metabolic-immune crosstalk driving immunotherapy resistance in CCA and propose a rationally designed combinatorial regimen to overcome therapeutic limitations, and offer an approach for enhancing clinical outcomes in this aggressive malignancy. Graphical model illustrating the mechanism by which USP1 promotes CCA progression and immune evasion. ACLY, glutathione S-transferase; ATG3, autophagy related 3; CCA, cholangiocarcinoma; CCL, C-C motif chemokine ligand; CCR5, C-C motif chemokine receptor 5; LD, lipid droplet; PGC1A, Peroxisome proliferator-activated receptor-γ coactivator-1α; PPAR, peroxisome proliferator activated receptor; TIME, Tumor immune microenvironment; USP1, ubiquitin-specific peptidase 1. Supplementary material Acknowledgements We would like to thank the Core Facility of the First Affiliated Hospital of Nanjing Medical University for its help in the detection of experimental samples. We thank Academic Journal Editing Services (AJE) (https://www.aje.cn/) for editing the English text of a draft of this manuscript Footnotes Funding: This work was supported by project grants from the Major Basic Research Fund of Jiangsu Province Hospital (QY202404), the National Natural Science Foundation of China (8227306, 82472865), the National Science Foundation for Young Scientists of China (82403149) and Jiangsu Provincial Basic Research Program (Natural Science Foundation, BK20220723). Provenance and peer review: Not commissioned; externally peer reviewed. Patient consent for publication: Not applicable. Ethics approval: The experiments on human participants were approved by Nanjing Medical University. Approval No. 2019-SR-133. Participants gave informed consent to participate in the study before taking part. Data availability statement Data are available upon reasonable request. All data relevant to the study are included in the article or uploaded as supplementary information. References Associated Data Supplementary Materials Data Availability Statement Data are available upon reasonable request. All data relevant to the study are included in the article or uploaded as supplementary information.