LOX-1 mediated lipid peroxidation leads to dysfunction of tumor infiltrating CD8⁺ T cells in ovarian cancer

Abstract

Abnormal lipid accumulation in the tumor microenvironment (TME) promotes lipid peroxidation in tumor-infiltrating CD8⁺T cells (CD8⁺TILs), driving their dysfunction and impairing antitumor immunity in ovarian cancer (OC). To elucidate the underlying immunosuppressive mechanisms, we integrated clinical specimens, mouse models, and in-vitro validation. Notably, CD8⁺TILs from OC patients exhibited pronounced lipid peroxidation signatures. Elevated oxidized low-density lipoprotein (oxLDL) levels were detected in ovarian tumor interstitial fluid (TIF), tumor mass-derived supernatants, and tissue sections, correlating with accelerated peroxidation phenotype in CD8⁺TILs. Bulk RNA sequencing and experimental validation revealed LOX-1 as the dominant oxLDL receptor in CD8⁺TILs, where its expression was positively correlated with peroxidation levels and negatively linked to PRF1, GZMB, and IFNG expression. In ID8 mouse OC models, LOX-1 expression and lipid peroxidation in CD8⁺TILs increased with tumor progression, and LOX-1-neutralizing antibody treatment partially reversed these changes and restored T-cell function. In vitro, LOX-1 blockade reduced oxLDL-induced lipid peroxidation, restored IFN-γ and GZMB production, and downregulated PD-1 expression, whereas LOX-1 overexpression exerted the opposite effects. Notably, the antioxidants, not ferroptosis inhibitors, restored CD8⁺T cell effector function by suppressing peroxidation. Mechanistically, oxLDL/LOX-1 signaling activated the transcription factor ELF3, which subsequently bound to the PDCD1 promoter to drive PD-1 expression. Together, this study demonstrates that in CD8⁺TILs, LOX-1-mediated uptake of oxLDL induced lipid peroxidation via oxidative stress and establish the oxLDL/LOX-1 axis as a pivotal pathway through which lipid peroxidation impairs CD8⁺TILs function in OC. Targeting LOX-1 or employing antioxidants restored the antitumor capacity of CD8⁺TILs, providing a mechanistic basis for metabolically modulating T cell-based cancer immunotherapy.

Graphical abstract

1. Introduction

While immune checkpoint inhibitors (ICIs) have achieved a huge success in treating several malignant cancers such as melanoma, its therapeutic efficacy remains suboptimal in ovarian cancer (OC), a disease widely regarded as a prototypical immunosuppressive malignancy [1]. A primary contributing factor is the limited infiltration and functional exhaustion of CD8⁺T cells within the highly immunosuppressive TME characteristic of ovarian cancer [2]. Therefore, developing strategies to effectively reinvigorate the suppressed CD8⁺TILs is both fundamentally and clinically important for treating OC and improving OC patient outcomes.

Intratumoral lipids have emerged as metabolic rheostats of T cell responses, with distinct species differentially modulating CD8⁺T cell fate and function [3]. OxLDL is a prominent oxidized lipid in the TME, and is present in free form within TIF [4]. Its increased uptake can disrupt cellular redox homeostasis and trigger lipid peroxidation in TILs, and multiple oxLDL receptor-mediated downstream signaling can further compromise CD8⁺T cell effector function, leading to a defective antitumor immunity [[5], [6], [7]]. Clinical data have indicated a positive correlation between elevated oxLDL levels and ovarian cancer progression [8]. However, existing studies have primarily focused on tumor cell-intrinsic metabolic alterations induced by oxLDL, its immunoregulatory role in shaping the tumor immune microenvironment, particularly its impact on TILs, remains poorly understood.

A critical and unresolved question is whether and how oxLDL-triggered lipid peroxidation serves as the key mechanistic link underlying CD8⁺TIL dysfunction. As a promising immunometabolic checkpoint in tumor immunity, studies in melanoma and colorectal cancer models demonstrated that sustained lipid peroxidation paradoxically fuels tumor progression by both suppressing anti-tumor CD8⁺T cells and programming Treg cells to adapt to a lactic acid-enriched TME, creating an immunosuppressive niche that promotes metastasis [[9], [10], [11]]. This oxidative stress could promote T cell exhaustion, marked by upregulated expression of PD-1 and TIM-3, while inducing ferroptosis further compromising antitumor immunity. Chu et al. recently reported that hepatocellular carcinoma (HCC) cells exploited this axis by overexpressing the cholesterol synthesis enzyme SQLE, catalyzing the production of large amounts of oxLDL [12]. Through paracrine signaling, these oxLDL molecules could recruit and polarize TREM2⁺macrophages, thereby suppressing CD8⁺TILs via metabolic-immune crosstalk, fostering an immune-evasive niche. Intriguingly, ovarian cancer exhibits unique lipid metabolic features. Hwang et al. found that the ovarian cancer TME further disrupted CD8⁺TILs function by suppressing TAGLN2, a key regulator of fatty acid transporter FABP5 localization, thereby starving anti-tumor CD8⁺T cells of functional lipid substrates despite abundant extracellular lipids [13]. These ovarian cancer-specific lipid metabolism profiles highlight critical gaps in understanding how the TME orchestrates lipid peroxidation to subvert CD8⁺TILs effector functions in ovarian cancer. Therefore, revealing the molecular mechanisms of how peroxidation impacts CD8⁺TILs will allow more effective therapeutics against ovarian cancer.

To address these gaps, we validated lipid peroxidation of CD8⁺TIL using clinical specimens, mouse models, and in-vitro validation, to determine whether and how oxLDL induces lipid peroxidation within CD8⁺TILs in ovarian cancer. Here, for the first time, we demonstrated that oxLDL suppressed effector function of CD8⁺TILs by enhanced peroxidation in ovarian cancer. Notably, LOX-1, one of the oxLDL receptors, is the prominent receptor mediating the peroxidation that impairs CD8⁺TILs in the TME. Mechanistically, LOX-1-mediated redox imbalance contributes to ELF3 upregulated PD-1 expression. Therapeutically, the administration of LOX-1 neutralizing antibodies elicited robust anti-tumor effects of CD8⁺TILs, indicating a promising target for ovarian cancer immunotherapy. This study elucidates ovarian cancer-specific immuno-metabolic suppression mechanisms and reveals actionable targets to restore CD8⁺TILs function in this ICIs treatment-insensitive malignancy.

2. Experimental model and study participation details

2.1. Human specimens

Peripheral blood samples from patients with OC and healthy donors during routine health check-ups were collected at the First Affiliated Hospital with Nanjing Medical University. Tumor tissues were obtained at the Affiliated Obstetrics and Gynaecology Hospital of Nanjing Medical University. Study protocols were reviewed and approved by the Institutional Review Board of the First Affiliated Hospital with Nanjing Medical University (Ethics Review No: 2023-SR-186) and the Affiliated Obstetrics and Gynaecology Hospital of Nanjing Medical University (Ethics Review No: PJ-2025KY023-001). Study protocols were in accordance with the Declaration of Helsinki. All patients did not receive any anti-tumor therapy and healthy donors had normal routine blood test results, as well as normal biochemical and tumor marker levels.

2.2. Mice

All animal experiments were performed in compliance with ARRIVE guidelines and the National Institutes of Health Guide for the Care and Use of Laboratory Animals. The experimental protocol was approved by the Institutional Animal Care and Use Committee (IACUC) of Nanjing Medical University (Approval No. IACUC-2409055). All efforts were made to minimize animal suffering, including the use of appropriate anesthesia and euthanasia procedures. Female C57BL/6J mice (6-8 weeks old) were obtained from the Specific Pathogen-Free (SPF) Animal Research Center of Nanjing Medical University. All mice were bred, housed, and maintained under controlled SPF conditions with the following environmental parameters: 22 ± 2°C ambient temperature, 40-60% relative humidity, and a 12:12-h light-dark cycle. Standard laboratory chow and autoclaved water were provided throughout the experimental period.

2.3. Method details

2.3.1. Cell lines culture

ID8 mouse ovarian epithelial cancer cells from, C57BL/6J mice were purchased from iCell (Cat#m064; RRID: CVCL_IU14). B16–F10 melanoma cells (from C57BL/6J mice) was a gift from Yuqing Wu (Nanjing Medical University) (RRID: CVCL_ 0159). ID8 cells and B16 cells were maintained at 37°C in 5% CO₂ in DMEM medium supplemented with 10% heat-inactivated fetal bovine serum and 1% penicillin and streptomycin.

2.3.2. Primary human peripheral blood immune cells

Primary peripheral blood mononuclear cells (PBMCs) were isolated from the peripheral blood of patients and healthy donors from the First Affiliated Hospital with Nanjing Medical University. Briefly, PBMCs were isolated by Ficoll-Paque PLUS (Cat# LTS1077, TBDscience) by density gradient centrifugation (450 g, 20 min, Thermo Scientific).

2.3.3. TIF and serum collection

TIF was collected from tumors using a previously described approach [5]. Tumors tissue were briefly rinsed in PBS and blotted on filter paper. The tumors were then put onto 70 μm cell strainers affixed atop 50 mL conical tubes, and centrifuged for 10 min at 4°C at 100 g. TIF was then collected from the conical tube, frozen in liquid nitrogen and stored at 80°C until further analysis.

To obtain human serum, peripheral blood samples from ovarian cancer patients and healthy donors were collected and centrifuged at 800 g for 10 min at 4°C. Then separated serum was carefully transferred into a clean tube and stored at −80°C.

2.3.4. Magnetic-activated cell sorting (MACS) cell separation and T cell activation

Human CD8⁺T cells were magnetically enriched using anti-human CD8 MicroBead Kit, according to the manufacturer's instructions (Miltenyi Biotec, cat#130-045-201). Enriched CD8⁺T cell were stained with CD8 antibodies and detected by the flow cytometry. The purity of CD8⁺T cells was >95%. Enriched CD8⁺T cells were washed and resuspended in X-Vivo15 medium (Lonza) containing 5% human AB serum (Gemini) and 500 U/mL IL-2 (PeproTech). For activation, CD8⁺T cells (1 × 10⁶ cells/well) were plated in an anti-CD3 antibody (5 μg/mL, OKT3; BioLegend)-coated 24-well plate and soluble anti-CD28 antibody (3 μg/mL,CD28.2; BioLegend) were added. After 2 days, the activated CD8⁺T cells were used.

2.3.5. Tissue processing

Fresh human/mouse tumor specimens were aseptically minced into <2 mm³ fragments using sterile scalpels in a biological safety cabinet. Tissue fragments underwent enzymatic digestion in RPMI-1640 medium (Gibco) supplemented with 1 mg/mL collagenase type IV (Gibco), 0.5 mg/mL hyaluronidase (Millipore Sigma), 0.1 mg/mL DNase I (Millipore Sigma), 2 mM l-glutamine (Gibco) for 1 h at a 37°C shaking incubator (200 rpm). The tumor suspension was then filtered using a 70 μm cell strainer, subsequently centrifuging. Tumor mass supernant were collected by culturing obtained cell pellets for 72 h at 37°C in 5% CO₂ in RPMI 1640 medium supplemented with 10% FBS and 1% penicillin and streptomycin. Meanwhile, homogeneous cell suspensions were centrifuged over the Percoll density gradient (GE Healthcare) and separated by collecting the interface fractions between 30% and 70% Percoll for human tumor tissue, as well as 40% and 80% Percoll for mouse tumor tissue. Mononuclear cells were isolated from the interface.

The spleens were mechanically triturated over 40 μm cell strainer (Corning) in cold PBS containing 2% FBS (Gibco). RBCs in spleen were lysed with Red Blood Cell Lysis Buffer (Beyotime) for 5 min.

2.3.6. Lipid peroxidation probes measurement

Lipid peroxidation measurement was performed using ROS Assay Kit -Highly Sensitive DCFH-DA (Dojindo, cat#R252), Liperfluo (Dojindo, cat# L248) and BODIPY 581/591 C11 kit (Dojindo, cat#L267) according to the manufacturer's instructions. Briefly, CD8⁺T cells were incubated with lipid peroxidation probes at 37°C for 30 min. CD8⁺T cells were then washed and examined by flow cytometry or fluorescence microscopy imaging after incubation with antibodies. BODIPY 581/591 C11 level was shown by FITC fluorescence or by calculating the fluorescence ratio of PE to FITC. A decrease in the ratio represents the increased level of lipid peroxidation. The data were analyzed using the FlowJo 10 and 7.6.1 software.

2.3.7. In vitro culture with oxLDL/LDL and OxLDL uptake assessment

CD8⁺T cells were treated with or without oxLDL (50 μg/mL, cat#YB-002-1, Yiyuan Biotech) or LDL (50 μg/mL, cat#YB-001, Yiyuan Biotech) for 48 h. For OLR1-overexpressing CD8⁺T cells, oxLDL stimulation was performed for 12 h. OxLDL is prepared by Cu²⁺-oxidation of ultracentrifugation-purified human native LDL in phosphate-buffered saline (pH 7.4) at 37°C, with oxidation terminated by excess EDTA-Na₂. Each lot is analyzed by agarose gel electrophoresis to confirm the characteristic electrophoretic mobility shift relative to native LDL. For the experiment involving ferroptosis inhibitors and antioxidants, ferrostatin-1 (MedChemExpress, cat#HY-100579, 5 μM), deferoxamine (MedChemExpress, cat#HY-B0988, 100 μM), α-Tocopherol (MedChemExpress, cat#HY-NO683, 200 μM), apocynin (MedChemExpress, cat#498-02-2, 500 μM) were added with oxLDL (50 μg/mL) for 48 h. Following incubation, cells were collected for further analysis.

OxLDL uptake was measured by incubating CD8⁺T cells with DiI-oxLDL (25 μg/mL, YB-0010, Yiyuan Biotech) in PBS for 4 h at 37°C. Following incubation cells were washed, surface stained, and the samples were analyzed by flow cytometry.

2.3.8. Tumor supernatant culture system

Healthy donors’ PBMCs derived CD8⁺T cells were isolated and activated as described above. Activated CD8⁺T cells (5 × 10⁴ cells/well) were cultured with either complete RPMI-1640 medium or tumor mass supernatant using Transwell permeable supports (0.4 μm, Corning). After 72 h co-culture, cells were harvested and subjected to phenotypic analysis. For the experiment involving antibody blockage, oriticumab (Selleck, cat#A2386, 10 μg/mL) and anti-LOX-1 (R&D Systems, cat#MAB1798, 1 μg/mL) were added in tumor supernatant culture system. Following incubation, cells were collected for further analysis.

2.3.9. ELISA and BCA

Human serum and tissue-derived supernatants were measured using oxLDL ELISA Kit (Sangon Biotech, cat#D711063) according to the manufacturer's instruction. The protein level of serum and TIF was measured using BCA Kit (Beyotime Biotechnology, cat#P0399S) according to the manufacturer's instruction.

2.3.10. RNA preparation and quantitative real-time PCR

Purified total RNA from sorted mouse/human tumor-infitrating CD8⁺T cells and CD8⁺T cells from mouse/human PBMCs were extracted using TRIzol (Invitrogen) and reverse transcribed with HiScript II Q Select RT Superix for qPCR (Vazyme, cat# R323-01). Real-time PCR amplifications were performed on the ABI 7500 Thermocycler (Applied Biosystems) in 20 μl reaction volumes containing complementary DNA, primers, and AceQ aPCR SYBR Green Master Mix (Vazyme, cat# Q711). The expression of a single gene was standardized to the expression of β-actin.

2.3.11. Bulk RNA-Seq and data analysis

CD8⁺T cells were treated with or without oxLDL. For RNA-seq, CD8⁺T cells were isolated from the PBMCs of healthy donors by using CD8 microbeads. RNA isolation, library construction, and sequencing were performed on a BGlSEQ-500 [Beijing Genomics Institute (BGI)]. Clean reads were mapped to the Homo sapiens genome assembly (GCF_000001405.40 GRCh38.p14) by HISAT2. For gene expression analysis, the mapped reads were calculated and then normalized to fragments per kilobase of transcript per million mapped reads. GO and KEGG pathway enrichment analyses were performed using R package based on the significantly differentially expressed genes with an adjusted p < 0.05.

2.3.12. Flow cytometry analysis

Cells were stained with the Fixable Viability Dye eFluor™ 780 (Thermo Fisher Scientific) to discriminate between live and dead cells, and then incubated with the surface antibodies for 30 min on ice. For staining of intracellular or endonuclear protein, cells were incubated with fixable viability dye prior to fixation and permeabilization (Invitrogen, Thermo Fisher Scientific) as per the manufacturer's instructions. After a washing step, cells were collected on a flow cytometer (Beckman Coulter Cytoflex S) and analyzed with FlowJoV10 software.

To stain cytokines, cells were first incubated for 4 h with 50 ng/mL PMA (Sigma-Aldrich), 1 μg/mL ionomycin (Sigma-Aldrich) in the presence of 10 μg/mL brefeldin (Sigma-Aldrich) at 37°C in complete RPMI 1640 medium.

Imaging flow cytometry of CD8⁺T cells treated with DiO (Servicebio, cat#G1704) and DiI-oxLDL were performed using ImageStream system (Amnis, USA) according to the manufacturer's instructions, and the data were analyzed using IDEAS 3.0.

2.3.13. Immunohistochemistry and immunofuorescence

Ovarian tissues were fixed in 4% PFA, then routinely paraffin-embedded and sectioned. Before staining, sections underwent heat-induced antigen retrieval. After blocking, they were incubated with ox-LDL polyclonal antibody (Bioss, cat#bs-1698R) or anti-CD8A monoclonal antibody (Abcam, cat#ab17147) for immunohistochemical staining, followed by HRP-conjugated secondary antibodies. Antigen-antibody complexes were visualized using DAB, and images were captured with a Nikon 50i microscope.

For immunofuorescence staining, tissue sections were incubated at 4°C overnight with anti-CD8A monoclonal antibody (Abcam, cat#ab17147 and cat#ab217344) or anti-pan cytokeratin monoclonal antibody (Abcam, cat#ab7753) or anti-LOX1 polyclonal antibody (Abcam, cat#ab126538). Slides were then incubated with indicated secondary antibodies. The nuclei were counterstained with DAPI. Slides were dried and mounted using ProLong Gold Anti-fade Mountant (Invitrogen, cat#P10144). Slides were visualized using a LEICA Stellaris STED microscope.

2.3.14. Tumor transplantation and in vivo treatments

In subcutaneous mouse tumor models, 5 × 10⁶ ID8 cells or 1 × 10⁶ B16 cells were engrafted per mouse. Tumor growth was assessed every week. Tumor volume was calculated using volume = 0.5 × length × width × height. For the i.p. model, 5 × 10⁶ ID8 cells were injected intraperitoneally. Ascites was collected for flow cytometry at 8 weeks. To block LOX-1 in mice, 1 μg per mouse anti-LOX-1 (R&D System, cat#AF1564) were injected i.p. every week or three days.

2.3.15. Cell depletion

To pharmaceutically deplete CD8⁺T cells in mice, the mice were treated with anti-mouse CD8α (Selleck, cat#A2102, Clone:2.43) antibodies 14 days before tumor inoculation. Anti-mouse CD8 were injected i.p. weekly each at 200 mg per mouse.

2.3.16. Lentiviral-mediated overexpression of LOX-1 and ELF3 in CD8⁺ T cells

For LOX-1 or ELF3 overexpression, activated CD8⁺T cells were transduced with lentiviral particles encoding human OLR1 or ELF3 (Genepharma), respectively, at a multiplicity of infection (MOI) of 100 in the presence of 6 μg/mL polybrene. After 24 h of transduction, the medium was replaced. Transduction efficiency was assessed by flow cytometry at 72 h post-transduction.

2.3.17. CRISPR-Cas9-mediated ELF3 knockout in CD8⁺ T cells

To knock out ELF3 in CD8⁺T cells, we employed the CRISPR-Cas9 system. Activated CD8⁺T cells were nucleofected with pre-assembled ribonucleoprotein (RNP) complexes consisting of recombinant Cas9 protein and synthetic guide RNA (sgRNA) targeting the human ELF3 gene (P1: 5′-GAGTACTGGGACTGTCTCGA-3′, P2: 5′-CGAAGACGCAGGTTCTGGAC-3′ General Bio), following the manufacturer's protocol (Nanoportal Biotech, cat#PT08). Knockout efficiency was confirmed by flow cytometry.

2.3.18. CUT&RUN

The CUT&RUN assay was conducted using Hyperactive pG-NNase CUT&RUN Assay Kit for qPCR (Vazyme, cat#HD101). Briefly, 10⁵ primary CD8⁺T cells were collected and washed once with 500 μL wash buffer before they were bound to ConA beads for 10 min at room temperature. After that, cells were incubated with 1 μg ELF3 antibody (CST) at 4°C overnight. Anti-mouse IgG was added and incubated for 1 h at 25°C the next day. Then cells were washed three times with Dig-wash buffer and incubated with 100 μL pGMNase Enzyme premix for 1 h at 4°C. Similarly, cells were washed three times with Dig-wash buffer, resuspended in fragmentation buffer and incubated at 37°C for 30 min. Add 100 μL of stop buffer to stop fragmentation and DNA was extracted by using column-based extraction reagents. DNA was eluted with double-distilled water. qPCR amplifications were performed on the ABI 7500 Thermocycler (Applied Biosystems) in 20 μl reaction volumes containing DNA, primers, and ChamQ Universal SYBR qPCR Master Mix. The expression of PDCD1 was calculated by a standard curve method and standardized to the expression of Spike in DNA. The sequences of PDCD1 primers were as follows: P1-S: 5′-GGGCGGGATATGGAAAGAGG-3′; P1-AS: 5′-GACAGAGGAGATGGGGAGGA-3′; P2-S: 5′-AGACACAGAGGAGGAAGGGG-3′; P2-AS: 5′-GGGGAGGGAGAGAGAGACAG-3′. The Spike in DNA sequence is as follows: ATAACTCAATGTTGGCCTGTATAGCTTCAGTGATTGCGATTCGCCTGTCTCTGCCTAATCCAAACTCTTTACCCGTCCTTGGGTCCCTGTAGCAGTAATATCCATTGTTTCTTATATAAAGGTTAGGGGGTAAATCCCGGCGCTCATGACTTCGCCTTCTTCCCATTTCTGATCCTCTTCAAAAGGCCACCTGTTACTGGTCGATTTAAGTCAACCTTTACCGCTGATTCGTGGAACAGATACTCTCTTCCATCCTTAACCGGAGGTGGGAATATCCTGCATTCCCGAACCCATCGACGA.

2.3.19. Luciferase assays

To assess the effect of ELF3 on PDCD1 promoter activity, 293T cells were transfected in two distinct setups: (1) pCDNA3.1-ELF3 + pGL3.1-PDCD1-WT + pRL-TK; (2) empty pCDNA3.1 + pGL3.1-PDCD1-WT + pRL-TK, using JETPRIME. After 48 h, cells were lysed and luciferase activity was quantified using the Dual Luciferase Reporter Assay System on a SpectraMax iD3 plate reader according to the manufacturer's instructions. Firefly luciferase signals were normalized to Renilla luciferase signals to calculate relative activity.

2.3.20. Statistical analysis

The data were analyzed by GraphPad Prism 9.0 software and are presented as the mean ± SEM. The statistics were analyzed by using an unpaired t test for two groups and a multiple t-test or two-way ANOVA for multiple groups. p values were provided as *p < 0.05; **p < 0.01; and ***p < 0.001.

3. Results

3.1. oxLDL enrichment in the TME induces lipid peroxidation of CD8⁺TILs

To validate the lipid peroxidation of CD8⁺TILs in ovarian cancer, we first analyzed bulk RNA-seq data of CD8⁺T cells from peripheral blood of healthy control donors (HCP), peripheral blood of ovarian cancer patients (OCP) and tumor tissues of ovarian cancer patients (OCT) (Fig. 1A). Results revealed significant enrichment of lipid peroxidation related pathways, such as oxidative stress, peroxidase activity and cellular oxidant detoxification, in CD8⁺T cells in OCT group (Fig. 1B and S Fig. 1A). Then, we assessed lipid peroxidation in CD8⁺T cells using multiple fluorescent probes. Flow cytometry analysis revealed that CD8⁺T cells from the OCT group exhibited significantly higher DCFH-DA level than those from both the HCP and OCP groups (Fig. 1C; representative fluorescence images shown in S Fig. 1B). In addition, CD8⁺T cells from the OCT group displayed markedly elevated Liperfluo and BODIPY 581/591 C11 signals compared with the OCP group (Fig. 1D and E; Liperfluo fluorescence images shown in S Fig. 1C). Collectively, these results indicate that CD8⁺TILs in ovarian cancer tissues undergo substantial lipid peroxidation. Meanwhile, in-vitro tumor supernatant culture system (Fig. 1F) was established to mimic TME. CD8⁺T cells in this system showed similar peroxidation phenotype, which was characterized by increased signals from all three florescent probes (Fig. 1G–I) and suppressed IFN-γ production (S Fig. 1D). Collectively, these data revealed that CD8⁺TILs exhibited a pronounced lipid peroxidation signature, suggesting a critical role for lipid peroxidation in driving T cell dysfunction in ovarian cancer.

Fig. 1.

CD8⁺TILs in the TME undergo increased lipid peroxidation due to uptake of oxLDL.

(A) Schematic of RNA sequencing performed on CD8⁺T cells using the following samples: peripheral blood from healthy donors (HCP), peripheral blood from ovarian cancer patients (OCP), and tumor tissues from ovarian cancer patients (OCT).

(B) The GO enrichment analysis and KEGG pathways analyses of differentially expressed genes (DEGs, P.adjust<0.05, |log2 fold change| >1) between OCT-CD8⁺T cells and OCP-CD8⁺T cells.

(C-E) DCFH-DA (C),Liperfluo (D) and BDP 581/591 C11 (E) level of CD8⁺TILs were analyzed by flow cytometry.

(F) Schematic of in-vitro tumor supernatant culture system in which CD8⁺T cells were treated with control medium (Con) or supernatant of tumor mass (Tu) for lipid peroxidation assessment using fluorescent probes.

(G-I) DCFH-DA (G),Liperfluo (H) and BDP 581/591 C11 (I) level of activated CD8⁺T cells in tumor supernatant culture system were assessed by flow cytometry.

(J) oxLDL concentration (pg/mL) in serum of healthy control donors (HC; n = 24), patients with benign ovarian tumor (BT; n = 24) and malignant ovarian tumor (MT; n = 29).

(K) oxLDL concentration (pg/mL) in supernatant of normal tissue adjacent to the tumor (NS; n = 7) and malignant ovarian tumor tissues (TS; n = 7).

(L) oxLDL/Protein (pg/mg) in serum (n = 32) and TIF (n = 24) of patients with malignant ovarian tumor.

(M) Immunohistochemical staining of oxLDL on the tissue sections of benign ovarian tumor (BT; n = 4), borderline ovarian tumor (BOT; n = 4) and malignant ovarian tumor (MT; n = 4).

(N) Immunofluorescence of oxLDL and CD8 on the tissue sections of malignant ovarian tumor (n = 5).

(O-Q) DCFH-DA (O),Liperfluo (P) and BDP 581/591 C11 (Q) level of activated CD8⁺T in tumor supernatants culture system in the presence of IgG or Orticumab.

(R) Liperfluo level of CD8⁺T cells treated with different concentrations of oxLDL (μg/mL) were analyzed by flow cytometry. (**p, ***p vs 0 group)

In the flow cytometry overlay histogram, the gray peak corresponds to the sample's FMO control. (C-E, G, H, O, P).

Data are presented as mean ± SEM. The p values were determined by unpaired two-tailed Student's t-test (C-E, G-I, K, R), paired t-test (O-Q), and Mann-Whitney U (J, L).

^nsp >0.05, *p < 0.05, **p < 0.01 and ***p < 0.001. ns, no significance.

Given the reported role of oxLDL in tumor promotion and its involvement in tumor oxidative lipid metabolism [5,12,14], we next measured the abundance of oxLDL in multiple types of OC clinical specimens. Quantitative analyses showed there was no difference in serum oxLDL among healthy donors (HC), benign ovarian tumor patients (BT) and malignant ovarian tumor patients (MT) (Fig. 1J). Crucially, OC tumor mass-derived supernatant (TS) exhibited greater oxLDL level than distant-normal tissues (NS) (Fig. 1K). Moreover, the oxLDL levels in TIF were much higher than serum in OC patients (Fig. 1L). We also found more oxLDL deposition in sections of malignant tumor tissue (MT) than benign (BT) and borderline tumors (BOT) (Fig. 1M). Importantly, co-localization of oxLDL with CD8 was observed and oxLDL accumulation in close proximity to CD8⁺TILs was also evident (Fig. 1N). Then, we observed CD8⁺TIL in ID8 subcutaneous mouse tumor models contained more endogenous oxLDL compared with CD8⁺T cells in peripheral blood (S Fig. 1E). To further confirm that oxLDL in tumor supernatants drives lipid peroxidation, we pre-incubated the supernatants with Orticumab, a specific anti-oxLDL neutralizing antibody. This treatment significantly reduced DCFH-DA, Liperfluo, and BODIPY 581/591 C11 signals in CD8⁺T cells (Fig. 1O–Q). Then, we found an oxLDL concentration-dependent lipid peroxidation increase in CD8⁺T cells (Fig. 1R). To exclude the possibility that the native LDL carrier itself accounts for the observed effects, we treated CD8⁺T cells with non-oxidized native LDL. Native LDL failed to elevate Liperfluo or BODIPY signals, whereas oxLDL induced robust lipid peroxidation (S Fig. 1F–H). A slight DCFH-DA elevation was observed in LDL group, likely reflecting a general response to lipid carriers.Thus, these results suggested that the lipid peroxidation in CD8⁺TILs in ovarian cancer is probably mainly driven by oxLDL, an important oxidative metabolite in the TME.

3.2. LOX-1 acts as the dominant oxLDL receptor on CD8⁺TILs and potentially drives tumor progression by inducing CD8⁺TILs functional impairment

To delineate oxLDL-uptake receptors on CD8⁺TILs, we first mined public ovarian cancer single-cell RNA-seq databases (http://ov.cancer-pku.cn/, GSE154600, Zhang et al.). Heatmap analysis revealed the expression of OLR1, CD68, MSR1, and SCARF1 was markedly upregulated in CD8⁺T cells from primary tumor compared to other tissue (Fig. 2A). We next validated these findings in our RNA-seq data. Volcano plot analysis of RNA-seq data from different CD8⁺T cell populations demonstrated that OLR1 was the most markedly altered oxLDL scavenger receptor gene in CD8⁺TILs (Fig. 2B–S Fig. 2A). Relative quantitative PCR demonstrated that OLR1 and MSR1 mRNA levels were elevated in CD8⁺TILs, with OLR1 showing the most significant upregulation (Fig. 2C and D). We then found OCT group show higher LOX-1 expression in CD8⁺T cells when compared with HCP and OCP groups (Fig. 2E). Gating strategy for LOX-1⁺CD8⁺T cells from tumor tissue has been shown in S Fig. 2B. Next, co-localization of LOX-1 and CD8A were observed on ovarian cancer sections (Fig. 2F), further suggesting LOX-1 as the primary oxLDL receptor on CD8⁺TILs. Furthermore, culturing with tumor mass-derived supernatants significantly upregulated LOX-1 expression in CD8⁺T cells (Fig. 2G). Gating strategy for LOX-1⁺CD8⁺T cells from tumor supernatant culture system has been shown in S Fig. 2C. Thus, these results indicated that LOX-1 may be the key receptor mediating the uptake of oxLDL in CD8⁺TILs. To further validate LOX-1 function, we examined DiI-labeled oxLDL (DiI-oxLDL) uptake in CD8⁺TILs and in the tumor supernatant culture system. In both contexts, LOX-1 neutralizing antibody (anti-LOX-1) significantly reduced oxLDL uptake (Fig. 2H and I). Then, we observed oxLDL highly co-localized with LOX-1 in CD8⁺T cells from in-vitro tumor supernatant culture system using Imaging flow (Fig. 2J). To further confirm the role of LOX-1 in oxLDL uptake, we overexpressed OLR1 in CD8⁺T cells. Imaging flow cytometry confirmed that LOX-1 protein was markedly increased on the cell surface after OLR1 overexpression (S Fig. 2D). Accordingly, CD8⁺T cells exhibited significantly enhanced DiI-oxLDL uptake upon direct oxLDL treatment in OLR1 overexpression group (Lv-OLR1) (Fig. 2K). Then, we found direct oxLDL exposure upregulated LOX-1 expression (Fig. 2L) and decreased cell viability (Fig. 2M) in a concentration-dependent manner in CD8⁺T cells. Thus, our results establish LOX-1 as the dominant functional receptor of CD8⁺TILs mediating oxLDL uptake in OC.

Fig. 2.

LOX-1 serves as the predominant receptor for oxLDL in CD8⁺TILs

(A) Gene expression of several oxLDL receptors in CD8⁺T cells from different sample types of ovarain cancer patients using public ovarian cancer single-cell RNA-seq databases (http://ov.cancer-pku.cn/).

(B) Volcano plot showing DEGs between OCT-CD8⁺T cells and OCP-CD8⁺T cells. Genes with a p-value <0.05 and |log2 fold change| > 1 are considered significant. Significantly up-regulated genes are shown in red, down-regulated genes in blue, and non-significant genes in gray. Several oxLDL scavenger receptors are marked with black arrows.

(C-D) Relative mRNA expression of oxLDL receptors in CD8⁺T cells from peripheral blood of healthy control donors (HCP), peripheral blood of ovarian cancer patients (OCP) and tumor tissues of ovarian cancer patients (OCT).

(E) Expression of LOX-1 on CD8⁺T cells surface was detected by flow cytometry.

(F) The co-localization of LOX-1 and CD8A on tissue sections of ovarian cancer was evaluated by immunofluorescence staining (n = 5).

(G) Expression of LOX-1 on CD8⁺T cells in in-vitro tumor supernatant culture system were analyzed by flow cytometry.

(H–I) DiI-oxLDL uptake by CD8⁺ T cells from ovarian cancer tissues (H) or tumor supernatant culture system (I) was measured by flow cytometry after treatment with IgG or anti-LOX-1.

(J) Schematic of detecting LOX-1 and oxLDL co-localization using exogenous DiI-oxLDL. The co-localization of LOX-1 and oxLDL in CD8⁺T cells were analyzed by Amnis ImageSteam flow cytometry. BF, bright field. The colocalization of LOX-1 and OxLDL was quantified based on the Bright Detail Similarity score computed by Amnis.

(K) CD8⁺T cells were transduced with lentivirus carrying empty vector (Lv-NC) or OLR1-overexpression vector (Lv-OLR1). Cells were then stimulated with DiI-oxLDL, and DiI-oxLDL uptake was assessed by flow cytometry.

(L) Expression of LOX-1 on CD8⁺T cells surface treated with different concentration of oxLDL (μg/mL) was measured by flow cytometry. (**p, ***p vs 0 group)

(M) Percentage of live cells in CD8⁺T cells treated with different concentration of oxLDL (μg/mL). (***p vs 0 group)

In the flow cytometry overlay histogram, the gray peak corresponds to the sample's FMO control (E, G-I, K).

Data are presented as mean ± SEM. The p values were determined by unpaired two-tailed Student's t-test (C-E, G, K-M) and paired t-test (H, I).

^nsp >0.05, *p < 0.05, **p < 0.01 and ***p < 0.001.

To explore the role of LOX-1 in ovarian tumor progression, we then mined public cancer gene expression and prognosis databases using web servers which analyze the data from The Cancer Genome Atlas (TCGA) database and published papers. The expression of OLR1 was significantly highly expressed in multiple tumors (S Fig. 3A). OLR1_low group of ovarian cancer patients showed better progression-free survival (PFS), overall survival (OS) and progressed-progression survival (PPS) (S Fig. 3B). Moreover, OLR1 exhibited high correlation with tumor infiltration of Tem_CD8⁺T cells in OC, preceded only by MDSC and macrophage (S Fig. 3C). We next investigated the relations of LOX-1 with both lipid peroxidation and the functional impairment of CD8⁺TILs. In clinical ovarian cancer specimens, we found that LOX-1⁺CD8⁺TILs exhibited substantially higher levels of lipid peroxidation compared with LOX-1⁻CD8⁺TILs (Fig. 3A–C). Identical trends were observed in tumor supernatant culture system (Fig. 3D and E). Meanwhile, in tumor supernatant culture system, LOX-1⁺CD8⁺T cell showed markedly reduced IFN-γ production, implying impaired anti-tumor function (Fig. 3F). Then, we stratified OC patients by OLR1 mRNA expression median. Results revealed that the LOX-1-high group displayed significantly lower mRNA levels of cytotoxic markers PRF1, GZMB and IFNG (Fig. 3G). Thus, these findings implicated LOX-1 potentially driving lipid peroxidation and impairing anti-tumor effector functions in OC CD8⁺TILs.

Fig. 3.

LOX-1 correlates with CD8⁺T cells cell peroxidation and function

(A-C) DCFH-DA (A), Liperfluo (B) and BDP 581/591 C11 (C) level of LOX-1⁺CD8⁺TILs and LOX-1⁻CD8⁺TILs were analyzed.

(D-E) DCFH-DA (D) and Liperfluo (E) level of LOX-1⁺CD8⁺T cells and LOX-1⁻CD8⁺T cells in tumor supernatant culture system were analyzed.

(F) Expression of IFN-γ of LOX-1⁺CD8⁺T cells and LOX-1⁻CD8+T cells in tumor supernatant culture system were analyzed.

(G) Relative mRNA expression of PRF1, GZMB and IFNG in CD8⁺TILs from LOX-1^low OC patients (L) and LOX-1^high OC patients (H) who stratified by OLR1 mRNA expression median were analyzed.

In the flow cytometry overlay histogram, the gray peak corresponds to the sample's FMO control (A, B, D, E).

Data are presented as mean ± SEM. The p values were determined by unpaired two-tailed Student's t-test (G) and paired t-test (A-F).

*p < 0.05, **p < 0.01 and ***p < 0.001.

3.3. Intervention of LOX-1 attenuated CD8⁺TILs lipid peroxidation and augmented antitumor cytokine secretion in mouse model

To further explore the role of LOX-1 in OC CD8⁺TILs lipid peroxidation and ovarian tumor progression, we then investigated the effect of LOX-1 during ovarian tumorigenesis. In the ID8 mouse model, compared with CD8⁺T cells in spleen and peripheral blood, CD8⁺TILs exhibited significantly elevated frequency of LOX-1, and this proportion further increased with progressive tumor growth (Fig. 4A and B). Concurrently, we observed substantially elevated DCFH-DA and Liperfluo levels within CD8⁺TILs at week 8 relative to week 4 (Fig. 4C and D). These data initially indicated the correlation between LOX-1 and lipid peroxidation in OC CD8⁺TILs. Then, administration of anti-LOX-1 during ID8 subcutaneous tumor modeling significantly inhibited tumor growth and increased the infiltration of CD8⁺TILs (Fig. 4E and F). To further validate this enhanced infiltration, we performed pan-cytokeratin/CD8A immunofluorescence (Fig. 4G) and CD8 immunohistochemistry on tumor sections (S Fig. 4A), which confirmed a marked increase in CD8⁺T-cell accumulation upon anti-LOX-1 treatment. The intervention of LOX-1 also suppressed levels of several lipid peroxidation probes in CD8⁺TILs (Fig. 4H–J) and restored CD8⁺TILs function, which were evidenced by significant upregulation of cytotoxic cytokines, including GZMB and IFN-γ (Fig. 4K and L). Meanwhile, we also observed similar results in B16 melanoma cancer model (S Fig. 4B–F). Also, in the i.p. model, anti-LOX-1 treatment recapitulated the key phenotypes observed subcutaneously, including increased ascitic CD8⁺T cell frequency and reduced DCFH-DA and Liperfluo levels (S Fig. 4G–I). Subsequently, to verify whether the anti-tumor effect of the anti-LOX-1 is specifically dependent on CD8⁺T cells, we performed an in-vivo CD8⁺T cell depletion experiment (S Fig. 4J). The gating strategy for CD8⁺T cells from ID8 tumor tissues is shown in S Fig. 4K, while the post-depletion CD8⁺T cell percentage in tumors is presented in S Fig. 4L. As expected, CD8⁺T cell depletion markedly attenuated the tumor-suppressive effect of anti-LOX-1 treatment, underscoring CD8⁺TILs as the primary mediators of this response. Intriguingly, a statistically significant difference in tumor growth between the anti-LOX-1 and IgG control groups appeared at week 7 and 8 post-engraftment (Fig. 4M). This residual effect suggests that LOX-1 blockade may engage additional CD8⁺T cell-independent mechanisms to inhibit tumor progression. Thus, these results collectively demonstrate that the involvement of LOX-1 in mediating lipid peroxidation in CD8⁺TILs and its blockade promote antitumor immunity by enhancing CD8⁺TILs function.

Fig. 4.

LOX-1 intervention alleviates lipid peroxidation and potentiates antitumor cytokine secretion in CD8⁺TILs in ID8 mouse model.

(A) Expression of LOX-1 in CD8⁺T cells from spleen, peripheral blood and subcutaneous tumor in 4-week and 8-week ID8 subcutaneous mouse models.

(B) Percentage of LOX-1⁺CD8⁺T cells in CD8⁺TILs in 4-week and 8-week ID8 subcutaneous tumor models.

(C) DCFH-DA level of CD8⁺TILs in 4-week and 8-week ID8 subcutaneous mouse models.

(D) Liperfluo level of CD8⁺TILs in 4-week and 8-week ID8 subcutaneous mouse models.

(E) Schematic of in-vivo anti-LOX-1 intervention using ID8 mouse models. Weight and growth of subcutaneous tumors in ID8 mouse models treated with anti-LOX-1 or IgG.

(F) Percentage and number of CD8⁺T cells infiltrated in subcutaneous tumor in ID8 mouse models treated with anti-LOX-1 or IgG.

(G)Immunofluorescence of pan-cytokeratin and CD8A on ID8 subcutaneous tumor sections (n = 4).

(H-J) DCFH-DA (H), Liperfluo (I) and BDP 581/591 C11 (J) level of CD8⁺TILs in ID8 subcutaneous mouse models treated with anti-LOX-1 or IgG.

(K-L) Expression of GZMB (K) and IFN-γ (L) in CD8⁺TILs in ID8 subcutaneous mouse models treated with anti-LOX-1 or IgG.

(M) Tumor growth curves in mice treated with anti-LOX-1 or control IgG or PBS (control) following CD8⁺T cell depletion (n = 4).

Data are presented as mean ± SEM. The p values were determined by unpaired two-tailed Student's t-test. Tumor growth curves were compared using two-way repeated-measures ANOVA.

*p < 0.05, **p < 0.01 and ***p < 0.001.

In the flow cytometry overlay histogram, the gray peak corresponds to the sample's FMO control (A, C, D, H, I).

3.4. LOX-1 serves as the key receptor mediating lipid peroxidation and anti-tumor function in CD8⁺TILs: evidence from in vitro models

We further demonstrated the crucial role of LOX-1 in regulating lipid peroxidation and antitumor immunity in CD8⁺TILs by performing LOX-1 neutralization and over-expression in vitro. Results showed that anti-LOX-1 significantly reduced fluorescence signal of the lipid peroxidation probes in CD8⁺T cells in tumor supernatant culture system (Fig. 5A–C). Similar results were observed in CD8⁺T cells which were directly exposure to oxLDL (Fig. 5D–E). Conversely, OLR1-overexpressing CD8⁺T cells exhibited significantly elevated Liperfluo levels following oxLDL stimulation compared to vehicle control (Fig. 5F). Additionally, this effect could be attenuated by anti-LOX-1 (S Fig. 5A). We then found that in oxLDL-stimulated CD8⁺T cells, anti-LOX-1 treatment significantly upregulated IFN-γ and GZMB levels (Fig. 5G–H), whereas OLR1-overexpression markedly downregulated IFN-γ and GZMB expression (Fig. 5I–J). Beyond cytotoxic antitumor cytokines, we also investigated immune checkpoints. Stratification of OC patients in the TCGA database into OLR1-high and OLR1-low cohorts revealed significantly elevated expression of multiple immune checkpoint molecules in the high-expression group (S Fig. 5B). Meanwhile, OLR1 exhibiting strong positive correlations with several checkpoints (S Fig. 5C). Next, we examined the effect of LOX-1 on immune checkpoints by OLR1 overexpression and oxLDL stimulation in CD8⁺T cells. Results revealed a selective upregulation of PDCD1 mRNA (Fig. 5K) after OLR1 over-expression. Meanwhile, elevated PDCD1 mRNA in CD8⁺T cells treated with oxLDL was observed when compared with control group (Fig. 5L). We further validated PD-1 upregulation at protein level after OLR1 over-expression (Fig. 5M). Intriguingly, anti-LOX-1 treatment notably blunted PD-1 upregulation in CD8⁺T cells in both the tumor supernatant culture system and upon oxLDL treatment (Fig. 5N and O). Collectively, through in-vivo animal models and in vitro cellular models, we concluded that LOX-1 is the pivotal receptor mediating oxLDL-induced lipid peroxidation and functional impairment in OC CD8⁺TILs. Blockade of LOX-1 effectively suppresses lipid peroxidation and functional impairment in CD8⁺TILs by increasing cytotoxic cytokines production and suppressing PD-1 expression.

Fig. 5.

LOX-1 mediates lipid peroxidation and antitumor function in CD8⁺T cells

(A-C) DCFH-DA(A), Liperfluo(B) and BDP 581/591 C11(C) level of CD8⁺T cells in tumor supernatant culture system were analyzed in the presence of anti-LOX-1 or IgG.

(D-E) Liperfluo(D) and BDP 581/591 C11(E) level of CD8⁺T cells treated with oxLDL were analyzed in the presence of anti-LOX-1 or IgG.

(F) Liperfluo level of CD8⁺T cells transduced with lentivirus carrying empty vector (Lv-NC) or OLR1-overexpression vector (Lv-OLR1) subsequently oxLDL treatment was assessed by flow cytometry.

(G-H) IFN-γ and GZMB expression of CD8⁺T cells treated with oxLDL were analyzed in the presence of anti-LOX-1 or IgG.

(I-J) IFN-γ and GZMB expression of CD8⁺T cells transduced with Lv-NC or Lv-OLR1 subsequently oxLDL treatment were analyzed by flow cytometry.

(K) Immune checkpoints relative mRNA expression of CD8⁺T cells transduced with Lv-NC or Lv-OLR1 subsequently oxLDL treatment were analyzed.

(L) Relative mRNA expression of PDCD1 of CD8⁺T cells with or without oxLDL treatment were analyzed.

(M) PD-1 expression of CD8⁺T cells transduced with Lv-NC or Lv-OLR1 were analyzed in the presence of oxLDL.

(N) PD-1 expression of CD8⁺T cells in tumor supernatant culture system were analyzed in the presence of anti-LOX-1 or IgG.

(O) PD-1 expression of CD8⁺T cells treated with oxLDL were analyzed in the presence of anti-LOX-1 or IgG.

In the flow cytometry overlay histogram, the gray peak corresponds to the sample's FMO control (A, B, D, F).

Data are presented as mean ± SEM. The p values were determined by unpaired two-tailed Student's t-test (D-M, O) and paired t-test (A-C, N).

^nsp >0.05, *p < 0.05, **p < 0.01 and ***p < 0.001.

3.5. CD8⁺TILs dysfunction induced by lipid peroxidation is attenuated by antioxidants but insensitive to ferroptosis inhibition

We next investigated molecular mechanisms underlying oxLDL-induced lipid peroxidation in CD8⁺T cells. Given documented roles of oxLDL in promoting oxidative damage across cell types, we evaluated the antioxidant compound vitamin E (α-tocopherol, Toco) and found that it significantly suppressed oxLDL-triggered lipid peroxidation (Fig. 6A–C), reversed inhibition of IFN-γ and GZMB expression, and reduced PD-1 expression (Fig. 6D–F). Consistently, Toco also reduced PD-1 expression in the tumor supernatant culture system (Fig. 6G). To extend these findings, we tested apocynin, a NADPH oxidase inhibitor that acts upstream of lipid peroxidation. Apocynin similarly reduced DCFH-DA, Liperfluo, and BODIPY signals (S Fig. 6A–C), increased IFN-γ and GZMB production and decreased PD-1 expression (S Fig. 6D–F). Together, these results indicate that antioxidants targeting distinct steps of the oxidative cascade can effectively reverse oxLDL-induced peroxidation and CD8⁺T cell dysfunction.

Fig. 6.

α-Tocopherol, but not ferroptosis inhibitors, reduces CD8⁺T cell lipid peroxidation and dysfunction.

(A-C) DCFH-DA(A), Liperfluo(B) and BDP 581/591 C11(C) level of CD8⁺T exposed to oxLDL with or without α-Tocopherol treatment were analyzed by flow cytometry.

(D-E) IFN-γ and GZMB expression of CD8⁺T cells exposed to oxLDL with or without α-Tocopherol treatment were analyzed by flow cytometry.

(F) PD-1 expression of CD8⁺T cells exposed to oxLDL with or without α-Tocopherol treatment were analyzed by flow cytometry.

(G) PD-1 expression of CD8⁺T cells in tumor supernatant culture system with or without α-Tocopherol treatment were analyzed.

(H) Liperfluo and BODIPY 581/591 C11 levels in CD8⁺T cells treated with 25 μg/mL oxLDL, with or without DFO or Fer-1, were analyzed by flow cytometry.

(I) Liperfluo and BODIPY 581/591 C11 levels in CD8⁺ T cells treated with 50 μg/mL oxLDL, with or without DFO or Fer-1, were analyzed by flow cytometry.

(J) Liperfluo and BODIPY 581/591 C11 levels in CD8⁺ T cells cultured with tumor supernatant, with or without DFO or Fer-1, were analyzed by flow cytometry.

In the flow cytometry overlay histogram, the gray peak corresponds to the sample's FMO control (A, B, H-J).

Data are presented as mean ± SEM. The p values were determined by unpaired two-tailed Student's t-test (A-F, H, I) and paired t-test (G, J).

^nsp >0.05, *p < 0.05, **p < 0.01 and ***p < 0.001.

Given that lipid peroxidation is the central biochemical event driving ferroptosis, and considering our observation of oxidative damage, we next investigated whether ferroptosis played a role in this process. Interestingly, neither ferrostatin-1 (Fer-1) nor deferoxamine (DFO) significantly attenuated lipid peroxidation in CD8⁺T cells treated with oxLDL at 25 or 50 μg/mL, or with tumor supernatant (Fig. 6H–J). Thus, our results indicated that the dysfunction of CD8⁺TILs triggered by oxidized lipids is mediated via a ferroptosis-independent pathway, which can be effectively mitigated by the antioxidants.

3.6. ELF3 impairs CD8⁺T cell anti-tumor function via oxLDL-induced PDCD1 activation

To further elucidate the molecular mechanism of oxLDL-induced lipid peroxidation in CD8⁺TILs, we systematically screened and validated key transcriptional regulators. Transcriptional factor (TF) enrichment analysis was performed on differentially expressed genes (DEGs) from three comparisons: OCT vs HCP, OCT vs OCP, and oxLDL vs control. To prioritize candidates, we integrated the results from the three KnockTF analyses with an additional TF enrichment analysis based on ChEA3 database. The intersection of these four enrichment analyses identified ten core TFs (Fig. 7A). Among these 10 TFs, ELF3 exhibited the most significant upregulation in the OCT group compared to the HCP and OCP groups (S Fig. 7A). Interesting, analysis of TCGA datasets confirmed ELF3 expression as the only significant predictor among these 10 TF of poor prognosis in OC patients (S Fig. 7B). Furthermore, in clinical specimens, ELF3 expression of CD8⁺T cells in OCT was significantly higher than CD8⁺T cells in HCP and OCP (Fig. 7B–C). Therefore, ELF3 emerges as a key TF of CD8⁺TILs functionally linked to TME and clinically associated with poor prognosis, prompting us to investigate its role in oxLDL-induced CD8⁺T cell dysfunction. Next, we found elevated ELF3 mRNA level in CD8⁺T cells which were directly stimulated with oxLDL (Fig. 7D). This ELF3 upregulation was abrogated by anti-LOX-1, amplified by LOX-1 over-expression, and attenuated by Toco co-treatment (Fig. 7E–G). These findings collectively implicated the involvement of ELF3 in oxLDL/LOX-1 mediated anti-tumor function impairment of CD8⁺T cells. We further analyzed the correlation between ELF3 and immune checkpoints to reveal the role of ELF3 in the downstream of oxLDL/LOX-1. Based on public ovarian cancer single-cell RNA-seq databases, results revealed statistically significant positive correlations between ELF3 and multiple immune checkpoint molecules (S Fig. 7C). Moreover, stratifying OC patients by ELF3 expression demonstrated significantly higher PDCD1 but comparable HAVCR2, LAG3, and CTLA4 mRNA levels in high-ELF3 group than low-ELF3 group (Fig. 7H). Consistent with clinical observations, ELF3 overexpression in CD8⁺T cells selectively upregulated PDCD1 mRNA without apparent changes in HAVCR2/LAG3/CTLA4 (Fig. 7I). To further validate the involvement of ELF3 overexpression in PDCD1 expression, we performed ELF3 knockout using CRISPR-Cas9 in CD8⁺T cells. Flow cytometric analysis revealed that, in the oxLDL direct stimulation model, PD-1 expression was markedly reduced in the ELF-3 knockout (ELF3-KO) group compared to the negative control (NC) group (Fig. 7J). Meanwhile, flow cytometric analysis confirmed enhanced PD-1 expression upon ELF3 overexpression (S Fig. 7D), collectively indicating that ELF3 specifically mediates oxLDL-driven PD-1 induction via transcriptional activation of PDCD1. Subsequent CUT&RUN assays validated direct ELF3 binding within the PDCD1 promoter region (Fig. 7K), while dual-luciferase reporter assays confirmed ELF3-dependent transcriptional activation of the PDCD1 promoter (Fig. 7L). Thus, these data demonstrate that ELF3 links the oxLDL/LOX-1 signaling axis to the PD-1, thereby revealing a novel lipid metabolism-phenotype regulation pathway that contributes to CD8⁺T cell dysfunction in ovarian cancer.

Fig. 7.

ELF3 impairs CD8⁺T cell anti-tumor function via oxLDL-induced PDCD1 activation

(A) Venn diagram illustrates the intersection of TFs significantly enriched among the DEGs (p-value <0.01 and |log2 fold change| > 3) from three comparisons: OCT vs HCP, OCT vs OCP, and oxLDL vs control. TF enrichment was performed using the KnockTF database for all three DEG sets and supplemented with ChEA3 analysis for the oxLDL vs control DEG set. Ten core TFs were identified at the intersection of the four enrichment analyses.

(B) Expression of LOX-1 on CD8⁺T cells surface was detected in HCP, OCP and OCT samples by flow cytometry.

(C) ELF3 relative mRNA expression of CD8⁺T cells in HCP, OCP and OCT samples were examined.

(D) ELF3 relative mRNA expression of CD8⁺T cells with or without oxLDL treatment were examined.

(E) ELF3 protein expression and mRNA expression of CD8⁺T cells treated with oxLDL were analyzed in the presence of anti-LOX-1 or IgG.

(F) ELF3 protein expression and relative mRNA expression were analyzed in CD8⁺T cells transduced with Lv-NC or Lv-OLR1 subsequently oxLDL treatment was assessed by flow cytometry.

(G) ELF3 protein expression and relative mRNA expression were analyzed in CD8⁺T cells exposed to oxLDL with or without α-Tocopherol treatment.

(H) Relative mRNA expression of PDCD1, HAVCR2, LAG3 and CTLA4 in CD8⁺TILs from ELF3^low OC patients (L) and ELF3^high OC patients (H) who stratified by ELF3 mRNA expression median.

(I) Relative mRNA expression of PDCD1, HAVCR2, LAG3 and CTLA4 in CD8⁺T cells transduced with lentivirus carrying empty vector (Lv-NC) or ELF3-overexpression vector (Lv-ELF3).

(J) PD1 protein expression of CD8⁺T cells in the ELF-3 knockout (ELF3-KO) group and negative control (NC) group was assessed by flow cytometry.

(K) CD3/CD28 activated CD8⁺T cells were treated with or without oxLDL for CUT&RUN-qPCR analysis of ELF3 binding at the PDCD1 gene locus (n = 3).

(L) The transcription activity of PDCD1 in 293T cells transduced with either the ELF3-overexpressing plasmid or an empty vector control were detected by dual-luciferase reporter assay.

In the flow cytometry overlay histogram, the gray peak corresponds to the sample's FMO control (B, E-G).

Data are presented as mean ± SEM. The p values were determined by an unpaired two-tailed Student's t-test.

^nsp >0.05, *p < 0.05, **p < 0.01 and ***p < 0.001.

4. Discussion

T cells in ovarian tumors are retained in a dysfunctional state characterized by bioenergetic anomalies, aberrant activation of cellular stress responses and negligible effector function that cannot be reversed through classical immunotherapeutic approaches [15]. Collective evidence strongly suggests that increased oxidative stress is a main driver of CD8⁺T cell dysfunction [16]. In the lipid overload tumor microenvironment, this equilibrium of oxidative stress is skewed toward an oxidative state via lipid peroxidation. This identifies lipid peroxidation as a key regulator of CD8⁺T cell-mediated antitumor immunity [3]. In this study, we hypothesize that enriched oxidized lipids in TME, such as oxLDLs, pose a peroxidation threat to CD8⁺TILs. Here, we identified a critical role of LOX-1 in promoting oxLDL-induced peroxidation of ovarian tumor CD8⁺TILs and supporting their potential to differentiate toward a dysfunction state via ELF3/PD-1 axis activation, thereby impairing CD8⁺T cell-mediated antitumor immunity and accelerating tumor growth. This work first illuminated immune-modulatory effects of oxidized lipids in ovarian cancer.

Dysregulated lipid metabolism is a hallmark of the TME, and increased oxLDL uptake and accumulation is observed in several types of intratumoral CD8⁺T cells, often associated with impaired anti-tumor immune function. Xu et al. observed significant differences in oxPLs (e.g. oxLDL) between Cd36^+/+ and Cd36^−/− CD8⁺TILs in B16 or MC38 tumors, demonstrating CD36 induced lipid peroxidation and CD8⁺T cell effector functional impairment in the involvement of p38 activation [5]. Qin et al. found CD36 promotes iron accumulation and lipid peroxidation in CD8⁺T cells by activating oxLDL-p38-CEBPB-TfR1 axis in early-stage hepatocellular carcinoma. Similarly, we demonstrated that oxLDL, which is enriched in the ovarian TME, drives lipid peroxidation in CD8⁺TILs and promotes their dysfunctional phenotype. However, there was no obvious elevation of CD36 on CD8⁺TILs in OC. Instead, we identified LOX-1 as playing the key role. Our findings revealed that LOX-1 mediates oxLDL uptake and peroxidation of CD8⁺TILs in OC, resulting in oxidative damage. This, in turn, dampens the antitumor response by decreasing cytokine production and activating the ELF3/PD-1 axis. Multiple receptors (LOX-1, CD36, TREM2, CXCL16) have been reported to recognize and internalize oxLDL in the TME, facilitating lipid accumulation, promoting pro-tumorigenic phenotypes in immune cells, and correlating with poor therapeutic response [5,7,12,17]. In this study, through receptor screening, co-expression analysis, overexpression assays, and neutralizing antibody blockade, our data strongly confirm that LOX-1 is the receptor dominating oxLDL uptake of CD8⁺T cells in the OC TME. The difference observed compared to previous reports may arise from the unique metabolic characteristics of ovarian tumors and the distinct expression patterns of scavenger receptors on CD8⁺TILs in OC. While our findings pinpoint LOX-1 as a critical regulator of CD8⁺TIL dysfunction in ovarian cancer, its role in tumor biology appears to extend beyond CD8⁺T cell immunomodulation. The observation that a significant attenuation, yet not a complete abolition, of tumor suppression occurred upon CD8⁺T cell depletion is particularly intriguing. While our data firmly identified CD8⁺TILs as the primary mediators of the anti-LOX-1 response, they also suggest the engagement of complementary, CD8⁺T cell-independent pathways. One compelling explanation lies in the potential cell-intrinsic role of LOX-1 within tumor cells themselves. Beyond its immunoregulatory function on T cells, LOX-1 is documented to be aberrantly expressed on various cancer cell types, including ovarian cancer, where it acts as a pro-tumorigenic receptor. Literature indicates that oxLDL/LOX-1 axis triggers an epithelial mesenchymal transition in prostate cancer cells and oxLDL/LOX-1 axis in tumor endothelial cells may lead to the formation of a high metastatic-tumor microenvironment via attracting neutrophils [18,19]. Meanwhile, LOX-1 was also reported to mediate the oxLDL uptake of pro-tumor neutrophils which displayed a higher potential to promote liver invasion [17]. Furthermore, studies revealed that OLR1 upregulation indicated poor prognosis in breast cancer and HNSCC, possibly through inducing macrophage polarization and triggering immune evasion [20,21]. This multifaceted role positions LOX-1 as a promising therapeutic target whose blockade could disrupt both the immunosuppressive tumor microenvironment and the autonomous growth capacity of cancer cells.

In recent years, along with the increasing attention to lipid peroxidation, ferroptosis, a type of programmed cell death initiated by overwhelmed lipid hydroperoxides in cells, has attracted great attention [22]. Ma et al. demonstrated that fatty acids induced ferroptosis in CD8⁺T cell lead to antitumor function impairment by increasing lipid peroxidation [23]. Meanwhile, Xu et al. and Qin et al. respectively found that activating NRF2 or overexpressing GPX4 reduces oxLDL-induced lipid peroxidation and restores CD8⁺T cell function [5,6]. Consistently, Srivastava et al. and Guo et al. reported that oxLDL induces ferroptosis in macrophages and AML cells [7,24]. These findings collectively suggest that targeting dysregulated lipid peroxidation and associated ferroptosis in CD8⁺TILs could enhance antitumor immune responses. In our study, we found antioxidants (Toco and apocynin) reversed the lipid peroxidation and suppression of antitumor function induced by both tumor mass-derived supernatant and oxLDL in CD8⁺T cells. However, we did not observe significant reductions in lipid peroxidation when using the widely employed ferroptosis inhibitors Fer-1 or DFO. Beyond cell-type specificity, one potential explanation for these discrepant findings is that Xu et al. demonstrated GPX4 overexpression rescued oxLDL-induced lipid peroxidation and decreased TNF/IFN-γ expression but did not use ferroptosis inhibitors to definitively confirm ferroptosis as the mechanism [22]. Moreover, although Qin et al. utilized Fer-1, their conclusion regarding oxLDL-induced ferroptosis in CD8⁺T cells was limited by the use of only two biological replicates and the measurement of a single lipid peroxidation marker, which is insufficient for reliable validation [6]. Another key possibility is the complex interplay between lipid peroxidation pathways and compensatory antioxidant mechanisms within cells, as ferroptosis execution requires not only lipid peroxidation accumulation but also the failure of cellular defense systems like GPX4 and NRF2, which is potentially masked in our tumor mass supernatant coculture system by the robust antioxidant activity of healthy donor PBMC-derived CD8⁺T cells.

Mechanistically, we found that oxLDL activated ELF3 regulation in a LOX-1-dependent manner and that suppression of lipid peroxidation via antioxidant diminished oxLDL-induced ELF3 expression. We also proved that ELF3 modulates PD-1 expression in CD8⁺T cells, as demonstrated by knockdown and overexpression experiments. Recently, PD-1 signaling was reported to inhibit phospholipid phosphatase 1 in CD8⁺TILs, reducing phosphatidylcholine and phosphatidylethanolamine synthesis and sensitizing T cells to unsaturated fatty acid-induced ferroptosis and impaired antitumor function [25]. Our work expands on this prior work by identifying actual upstream triggers and transcription factors in the TME that up-regulate PD-1 expression in TILs and highlighting the role of oxidative stress responses in CD8⁺T cell dysfunction in the context of ovarian cancer.

In summary, our study indicates that targeting LOX-1 could be an effective strategy to improve the antitumor efficacy of T cell-based immunotherapy in ovarian cancer. Our study also provides a possible approach to metabolically manipulate CD8⁺T cells to improve the clinical effectiveness of cancer immunotherapies.

Consent for publication

All authors have read and approved the final version of the manuscript. We confirm that this work is original, has not been published previously, and is not under consideration for publication elsewhere.

Availability of data and materials

All the data generated and utilized in this study are included in this published article.

Ethics approval and consent to participate

Studies involving human participants were reviewed and approved by the Ethical Committee of the First Affiliated Hospital of Nanjing Medical University (Nanjing, China) (Ethics Review No: 2023-SR-186) and the Affiliated Obstetrics and Gynaecology Hospital of Nanjing Medical University (Ethics Review No: PJ-2025KY023-001). All animal experiments complied with institutional guidelines and were approved by the Animal Welfare and Ethics Committee of Nanjing Medical University (IACUC-2409055).

U.Funding statement

This work was supported by following grants: National Natural Science Foundation of China (no. 82502175 to Ting Wang, no. 82273199 to Fang Wang, no. 82202610 to Yuexinzi Jin), Jiangsu High-level Hospital Pairing Assistance Research Initiative (no. JDBFSN202503 to Tingting Song), Joint special project of Baoshan medical research (no. 2024bskylms004 to Xin Fu).

CRediT authorship contribution statement

Ting Wang: Conceptualization, Data curation, Formal analysis, Funding acquisition, Writing – original draft. Lina Yan: Data curation, Formal analysis, Methodology, Software. Xi Huang: Data curation, Formal analysis, Investigation, Methodology. Lingfei Zhou: Data curation, Project administration, Software. Yepeng Mao: Data curation, Investigation, Methodology. Shuna Liu: Investigation, Methodology, Software. Xin Fu: Funding acquisition. Jinghang Li: Investigation, Methodology. Ziqi Tao: Methodology, Software. Zhijie Liu: Methodology, Software. Yuelu Zhang: Methodology, Validation. TingTing Song: Funding acquisition. Yuexinzi Jin: Funding acquisition, Project administration, Writing – review & editing. Fang Wang: Conceptualization, Resources, Supervision, Writing – review & editing.

Declaration of competing interest

The authors declare the following financial interests/personal relationships which may be considered as potential competing interests:Ting Wang reports financial support was provided by National Natural Science Foundation of China. Fang Wang reports financial support was provided by National Natural Science Foundation of China. Yuexinzi Jin reports financial support was provided by National Natural Science Foundation of China. Tingting Song reports financial support was provided by Jiangsu High-level Hospital Pairing Assistance Research Initiative. Xin Fu reports financial support was provided by Joint special project of Baoshan medical research. If there are other authors, they declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Appendix A. Supplementary data

The following is the supplementary data to this article:

Data availability

Data will be made available on request.