Clinical insight-driven micron-sized cholesterol oxidation platform for membrane lipid therapy of advanced ovarian cancer

Abstract

Due to the insidious onset of ovarian cancer, the majority of patients are diagnosed at advanced stages, often presenting with extensive pelvic and abdominal metastasis. A significant proportion of these patients exhibit primary or acquired drug resistance, limiting improvements in the five-year survival rate. In this study, an analysis of patient-derived samples demonstrated a strong correlation between cholesterol levels and ovarian cancer progression. Membrane lipid therapy, a strategy that targets the composition, structure, and function of the cell membrane, has demonstrated notable potential in tumor therapy. Subsequently, a novel cholesterol oxidation-mediated membrane lipid therapy employing PLGA microspheres co-loaded with miriplatin (MiR) and cholesterol oxidase (COD) was proposed for drug-resistant advanced ovarian cancer therapy. The microspheres had a particle size of approximately 2.58 μm, with encapsulation efficiencies of 82.98 ± 0.09% for MiR and 32.83 ± 4.30% for COD. In vitro experiments demonstrated that COD-induced cholesterol oxidation modulated the membrane rigidity and fluidity of SKOV3-TR cells, thereby compromising membrane structural integrity and attenuating tumor cell migration. Additionally, the reactive oxygen species generated during cholesterol oxidation disrupted mitochondrial membrane potential and adenosine triphosphate production. The resulting energy deficiency and compromised membrane integrity reduced the expression and function of drug resistance-associated proteins, thereby enhancing chemosensitivity. Moreover, the combined effects of reactive oxygen species and MiR drove resistant cells towards apoptosis. In vivo studies demonstrated that the large-particle PLGA formulation effectively resided in the peritoneal cavity, resulting in superior therapeutic outcomes against drug-resistant metastatic ovarian tumor, as evidenced by fewer peritoneal metastatic nodules and diaphragmatic colonization sites, more extensive tumor tissue destruction, and prolonged survival. More importantly, PLGA encapsulation significantly decreased the toxic side effects associated with continuous platinum-based chemotherapy. Overall, cholesterol oxidation-mediated membrane lipid therapy represented a promising approach for treating advanced ovarian cancer.

Graphical abstract

Supplementary Information

The online version contains supplementary material available at 10.1186/s12951-025-03701-7.

Introduction

Ovarian cancer (OC) is a highly aggressive malignancy that is often diagnosed at advanced stages, frequently accompanied by widespread pelvic and abdominal metastasis. This complexity greatly hinders surgical resection, positioning chemotherapy as the primary treatment option for patients with advanced ovarian cancer. The standard first-line treatment typically involves a combination of paclitaxel and platinum-based chemotherapy. Unfortunately, approximately 20% of OC patients exhibit inherent resistance to this chemotherapy regimen, while about 80% develop secondary resistance during the course of treatment [1]. The high incidence of metastasis and drug resistance in advanced OC are key factors leading to treatment failure [2]. Although maintenance therapy, particularly with poly (ADP-ribose) polymerase inhibitors, and subsequent treatments have been integrated into the management of intermediate and advanced OC, global cancer statistics from 2024 demonstrated that the five-year survival rate for OC has remained stagnant at 30–50% over the past several decades [2]. Therefore, developing new therapies for advanced ovarian cancer is an urgent challenge and key focus in gynecologic oncology.

Recent studies have demonstrated that the plasma membrane functions not only as a barrier to drug delivery but also as a critical target in cancer therapy [3, 4]. Previously, our team proposed “membrane lipid therapy”—a strategy aimed at modulating membrane composition, structure, and function [5]. Drug-resistant tumor cells exhibit distinct lipid compositions and membrane characteristics compared to sensitive cells, thereby reducing the damage from chemotherapy drugs [6, 7]. Notably, the cholesterol content in the membranes of drug-resistant cells is considerably elevated compared to that in sensitive cells [8]. Platinum-resistant OC cells increase their extracellular cholesterol uptake by up-regulating receptors [9]. Additionally, these resistant cells improve cholesterol synthesis by activating the mevalonate pathway and upregulating sterol regulatory element-binding protein 2, among others [10, 11]. The Cancer Genome Atlas demonstrated a significant correlation between the overexpression of genes involved in cholesterol biosynthesis and poor survival outcomes in tumor patients [12]. Several studies showed that the increased cholesterol content in the membrane of drug-resistant tumor cells enhanced membrane stiffness, which in turn impeded drug penetration and contributed to the development of tumor drug resistance [13, 14]. Lipid rafts, membrane microdomains enriched in cholesterol and sphingomyelin, play a critical role in various cellular functions, such as endocytosis and vesicular transport [15]. An increase in lipid raft composition is strongly linked to the development of drug resistance in tumors [16, 17]. The elevated levels of cholesterol and sphingomyelin in lipid rafts of drug-resistant cancer cells facilitate the expression, recycling, and functional activity of multidrug resistance transporters, such as P-glycoprotein (p-gp), which accumulate in these domains [18, 19], accelerating the efflux of chemotherapeutic drugs [20].

Besides its association with tumor drug resistance, the metastatic potential of tumor cells is also closely linked to the physical properties of the plasma membrane [21, 22]. The cholesterol level in the plasma membrane is essential for modulating its fluidity, rigidity, and overall structural integrity [23, 24]. Metastatic tumor cells often dynamically remodel their membranes to adapt to different microenvironments and engage in invasive movements. Elevated cholesterol levels in the membrane enhance its rigidity, support the functionality of membrane proteins, alter cell shape, and facilitate the membrane remodeling that is essential for the migration of tumor cells [6]. Moreover, isoprene, a byproduct of cholesterol synthesis, serves as a cellular second messenger, stimulating adenosine triphosphate (ATP) biosynthesis in mitochondria. This process provides the necessary energy for the metastasis of tumor cells [25]. To explore novel therapeutic strategies for advanced ovarian cancer, this study analyzed patient samples to investigate the association between cholesterol levels and disease progression, based on which a cholesterol-targeted treatment strategy was proposed.

Cholesterol oxidase (COD) is a key enzyme involved in cholesterol metabolism and is widely utilized as a diagnostic agent for cholesterol detection in medical practice. It effectively catalyzes the selective oxidation of cholesterol, leading to the production of 4-cholesten-3-one and hydrogen peroxide (H₂O₂) [26]. Notably, COD-mediated cholesterol oxidation within cells reduces membrane cholesterol levels, thereby altering the physicochemical properties of the plasma membrane [27, 28]. Interestingly, the H₂O₂ produced during the cholesterol oxidation process can generate highly oxidative hydroxyl radicals (·OH) under the catalysis of endogenous Fenton-like catalysts (such as Fe²⁺/³⁺) [14]. The accumulation of intracellular ·OH impairs mitochondrial function and promotes apoptosis in cancer cells. Therefore, COD may serve as a potential molecular strategy to overcome platinum resistance and metastasis in advanced ovarian cancer.

As a natural enzyme, COD may pose potential risks to normal tissues. Polylactic-hydroxyacetic acid copolymer (PLGA) is extensively utilized as a drug delivery carrier in clinical formulations owing to its multiple advantages. Compared to smaller nanocarriers such as liposomes or hexosomes, PLGA microspheres offer advantages for delivering hydrophilic macromolecules like COD. The polymeric matrix of PLGA provides effective protection for labile protein drugs and enables controlled, sustained release for long-term therapeutic effects. Moreover, the particle size and surface characteristics of PLGA microspheres may promote better adhesion to the peritoneal surface after intraperitoneal (I.P.) administration, potentially enhancing local drug retention and therapeutic efficacy. Consequently, this study designed PLGA microspheres encapsulating miriplatin (MiR) and COD (PLGA@MiR&COD) to investigate its therapeutic efficacy and mechanism in inhibiting ovarian cancer metastasis and restoring the sensitivity of drug-resistant tumor cells to MiR. Specifically, a clinically relevant ovarian cancer model, recapitulating both metastatic behavior and chemoresistance, was established to assess the therapeutic potential of PLGA@MiR&COD. Furthermore, the potential toxic effects of the novel formulation were evaluated within the animal model.

Materials and methods

Materials

Cholesterol oxidase and miriplatin were obtained from Macklin Biochemical Co., Ltd. (Shanghai, China). Fetal bovine serum (FBS) was purchased from Thermo Fisher Scientific, Inc. (MA, USA). RPMI-1640 medium was obtained from Shanghai Yuanpei Biotechnology Co., Ltd. (Shanghai, China). Dilute hydrochloric acid solution and 2,7-dichlorofluorescein diacetate (DCFH-DA) were obtained from Shanghai Yuanye Biotechnology Co., Ltd. (Shanghai, China). Poly (D, L-lactide-co-glycolide) (MW 7,000–17,000, 50%:50%), platinum standard solution, Cyanine 5 (Cy5) fluorophore, 1-pyrenyldiacrylamide, and β-Cyclodextrin (β-CD) were purchased from Aladdin Biochemical Technology Co., Ltd. (Shanghai, China). Dichloromethane was obtained from Sinopharm Chemical Reagent Co., Ltd. (Shanghai, China). The poly (vinyl alcohol) (PVA) was obtained from Sigma-Aldrich, Merck KGaA (Missouri, USA). The ATP assay kit and cholesterol assay kit (Cell-Based) were purchased from Abcam plc (Cambridge, UK). The AF488-CTB fluorescent dye solution, JC-1 probe, and 4′,6-diamidino-2-phenylindole (DAPI) were obtained from Thermo Fisher Scientific, Inc. (MA, USA). The FITC Annexin V Apoptosis Detection Kit I and rhodamine 123 were purchased from Beyotime Biotechnology Co., Ltd. (Suzhou, China). The TRIzol reagent was obtained from Invitrogen, Thermo Fisher Scientific, Inc. (MA, USA). The TMR (Red) TUNEL Cell Apoptosis Detection Kit and reverse transcription kit were purchased from Servicebio Technology Co., Ltd. (Wuhan, China). The CCK-8 assay kit was obtained from Meilun Biotechnology Co., Ltd. (Dalian, China). P-glycoprotein (p-gp) antibody for Western blotting was purchased from Abcam plc (Cambridge, UK).

Cells and animals

The SKOV3 human ovarian cancer cell line was sourced from the American Type Culture Collection (Manassas, USA). The drug-resistant SKOV3 cell line (SKOV3-TR) was established by continuously exposing parental SKOV3 cells to miriplatin over an extended period. The SKOV3-TR cells were cultured in RPMI-1640 medium, supplemented with 10% fetal bovine serum and 1% Penicillin-Streptomycin solution. Female BALB/c nude mice, aged 4 to 6 weeks, were purchased from Shanghai SLACK Laboratory Animal Co., Ltd. (Shanghai, China) and kept in the specific pathogen-free (SPF) facility at the Zhejiang University of Traditional Chinese Medicine Animal Experiment Center.

Collection of retrospective data from OC patients

This study analyzed the clinical data from 192 patients with epithelial ovarian cancer (EOC), previously included in our published research cohort [29]. Each patient underwent primary debulking surgery, followed by platinum-based chemotherapy. In addition to the clinical and pathological data from our previous study, preoperative and pre-chemotherapy serum total cholesterol levels (recorded before each chemotherapy cycle) were obtained from the medical records of the Women’s Hospital, Zhejiang University School of Medicine.

Determination of cholesterol levels in patient tumor tissues

In this study, 10 platinum-resistant and 10 platinum-sensitive patients were selected by a matched sampling approach based on age, International Federation of Gynecology and Obstetrics (FIGO) stage, and tumor grade, aiming to evaluate cholesterol levels in both serum and postoperative tumor tissues of ovarian cancer patients. Formalin-fixed and paraffin-embedded tissue samples from 20 EOC patients were subjected to Filipin fluorescence staining. The process of dewaxing and staining the samples was performed according to the reagent instructions, and the fluorescence intensity was subsequently examined under a fluorescence microscope (Leica, Germany). Fluorescence intensity of Filipin-stained sections was quantified using ImageJ software.

Preparation and characterization of drug-loaded microspheres

In this study, PLGA microspheres were prepared using a double emulsion method. The initial selection of the solvent, ultrasound power, and sonication time was guided by previously published protocols [30, 31]. Formulation optimization was subsequently performed by adjusting the input amounts of PLGA and miriplatin, with the selection criteria aiming for a particle size of approximately 2 μm and an encapsulation efficiency of MiR exceeding 80%. The final mass ratio of MiR, COD, and PLGA was determined to be 2:5:52. To prepare PLGA@MiR&COD, 2 mg of MiR and 52 mg of PLGA were dissolved in 600 µL and 2 mL of dichloromethane, respectively, and mixed thoroughly. Subsequently, 5 mg of COD dissolved in 200 µL of water was added. The mixture underwent probe sonication under ice bath conditions (150 W for 2 min). The resulting primary emulsion was added to 10 mL of 2% PVA solution and further sonicated (100 W for 3 min) to form a double emulsion. Then, 10 mL of 1% PVA was added, and dichloromethane was removed via rotary evaporation. The emulsion was centrifuged and washed three times with 1% PVA solution to remove free MiR and COD. The final PLGA@MiR&COD formulation was stored at 4°C for further use. To prepare PLGA@MiR, the procedure was identical, except that 200 µL of COD aqueous solution was replaced with 200 µL of ultrapure water. The particle size and zeta potential of PLGA@MiR and PLGA@MiR&COD were analyzed using a Malvern laser particle size analyzer (Malvern, USA). Stability assessment was conducted by dispersing PLGA@MiR&COD in PBS or PBS with 10% FBS at a concentration of 5 mg/mL. Particle size was measured at designated time points over 7 days to evaluate colloidal stability. Scanning electron microscopy (SEM, ZEISS, Germany) was employed to investigate the structure and morphology of PLGA@MiR&COD, while transmission electron microscopy (TEM, Hitachi H-7650, Japan) was utilized to further examine the morphological features of PLGA@MiR&COD.

Encapsulation efficiency determination and in vitro release study

The encapsulation efficiency of PLGA microspheres was determined using a centrifugal method. Following the absorption of the supernatant, the concentration of COD in the supernatant was measured using a protein concentration determination kit, and the concentration of MiR in the supernatant was quantified by an inductively coupled plasma spectrometer (Agilent, USA). Finally, the encapsulation rates of PLGA microspheres for MiR and COD were calculated according to the following formulas: encapsulation efficiency (%) = (mass of encapsulated drug/mass of drug initially added during preparation) × 100%.

The in vitro release study was conducted using the dialysis method. The freshly prepared PLGA@MiR&COD dispersed liquid was diluted with PBS solution to a total volume of 9 mL and then placed into a 10 mL dialysis tube (MWCO: 3500 Da). The tube was submerged in 500 mL of PBS at pH 7.4 with or without 10% FBS, or at pH 6.5 with 10% FBS, and incubated at 37°C with continuous shaking at 100 rpm. At predetermined time points (0.5, 1, 2, 4, 8, 12, 24, and 48 h), 1 mL samples were withdrawn, and an equal volume of fresh PBS was added to maintain the volume. The collected samples were diluted to a final volume of 5 mL with 0.1 M HCl solution and subsequently filtered through a membrane prior to analysis. Platinum concentrations were quantified using inductively coupled plasma mass spectrometry (ICP-MS, Thermo Fisher Scientific, USA). Calibration was conducted using a series of platinum standard solutions with concentrations of 0, 3.0, 6.0, 9.0, 15.0, 30.0, 60.0, and 120.0 µg/L. The resulting calibration curve demonstrated excellent linearity, with a correlation coefficient (R²) greater than 0.999. The detection limit for platinum was established at 0.01 µg/L. The release profile was then constructed based on these measurements.

Hemolysis assay

Fresh anticoagulated rat blood was centrifuged to remove plasma. The erythrocytes were washed three times with saline and resuspended to form a 2% (v/v) suspension. Samples were divided into five groups: pure water (positive control), saline (negative control), and test formulations at 25, 50, and 100 µg/mL. Each group was mixed with an equal volume of erythrocyte suspension and incubated at 37°C for 1 h. After centrifugation, the absorbance of the supernatant was measured at 545 nm. Hemolysis ratio = (OD_T – OD_N)/(OD_P – OD_N) × 100%, where OD_T, OD_N, and OD_P represent the optical densities of the test group, negative control, and positive control, respectively. A hemolysis rate below 5% was considered non-hemolytic.

Cellular uptake and intracellular localization study

The MiR was replaced with Cy5 to prepare Cy5-labeled PLGA microspheres (PLGA@Cy5). The SKOV3-TR cells were exposed to a culture medium with or without PLGA@Cy5 for 0.5, 1, 4, or 8 h. Fluorescence intensity was measured using flow cytometry (BD Biosciences, USA). The intracellular localization of PLGA@Cy5 was observed via confocal laser scanning microscopy (CLSM, Olympus, Japan). For CLSM analysis, SKOV3-TR cells were cultured overnight in a confocal dish and then treated with a medium containing PLGA@Cy5. After fixation with 4% paraformaldehyde (v/v), the nuclei were stained with DAPI before examination.

Cytotoxicity analysis

The cytotoxic effects of free MiR, free COD, the physical mixture of MiR and COD (MiR + COD), and PLGA@MiR&COD on SKOV3-TR cells were assessed using CCK-8 assays. The SKOV3-TR cells were exposed to free MiR, free COD, MiR + COD, or PLGA@MiR&COD for 24–48 h. After treatment, 15 µL of CCK-8 reagent (5 mg/mL) was added to each well, and absorbance was measured at 450 nm following a 2-hour incubation period.

Measurement of membrane cholesterol content

SKOV3-TR cells were exposed to various treatments, including drug-free medium, β-CD, free MiR, free COD, MiR + COD, or PLGA@MiR&COD-containing medium. The β-CD, widely recognized for its ability to deplete membrane cholesterol, was selected as a positive control in this study [32]. A concentration of 5 mM β-CD was fully dissolved in solution without any visible precipitation. Both COD and MiR were used at a concentration of 40 µg/mL. The β-CD was applied as a positive control for cholesterol complexation. Following a 24-hour incubation, the cells were incubated in the dark with Filipin III solution (50 µg/mL) for 2 h. The cholesterol localization and levels in cells of each treatment group were examined using the CLSM.

Determination of membrane lipid raft levels

SKOV3-TR cells were seeded in confocal chambers and cultured for 24 h. After that cells were treated with drug-free medium, free MiR, free COD, MiR + COD, or PLGA@MiR&COD, each containing MiR and/or COD. After 24 h of incubation, cells were then incubated with AF488-CTB (5 µg/mL) for 30 min, fixed with 4% paraformaldehyde for 15 min, and stained with DAPI for another 15 min. Membrane lipid raft levels were observed using CLSM.

Examination of plasma membrane fluidity

This study utilized the pyrene fluorescence method to measure the membrane fluidity of cells after different treatments. Following an overnight culture, the SKOV3-TR cells were exposed to either drug-free medium, free MiR, free COD, MiR + COD, or PLGA@MiR&COD-containing medium. After that, cells were cultured for an additional 24 h. Post-incubation, cells were subsequently incubated with 1-pyrenyldiacrylamide for 15 min. Membrane fluidity was determined based on the ratio of optical density values at 475 nm (aggregates) and 397 nm (monomers) using a Microplate Reader (Thermo Fisher Scientific, USA).

Quantification of p-gp mRNA and protein expression in SKOV3-TR cells

SKOV3-TR cells were treated with drug-free medium, free MiR, free COD, MiR + COD, or PLGA@MiR&COD-containing medium for 24 h. Total RNA and protein were extracted using TRIzol reagent and RIPA buffer, respectively. For mRNA analysis, cDNA was synthesized using a reverse transcription kit, and quantitative real-time PCR was performed on a Bio-Rad PCR system (California, USA). The p-gp mRNA levels were determined using specific primers (forward: 5’-GGTTGCTGCTTACATTCAGGTTT-3’; reverse: 5’-CAGCCTATCTCCTGTCGCATTA-3’), with GAPDH as the internal control (forward: 5’-GGAAGCTTGTCATCAATGGAAATC-3’; reverse: 5’-TGATGACCCTTTTGGCTCCC-3’). For protein analysis, p-gp expression was assessed by Western blot using a specific antibody, with β-actin as the loading control.

Measurement of intracellular levels of reactive oxygen species

SKOV3-TR cells were plated in confocal chambers. Reactive oxygen species (ROS) production was assessed using the DCFH-DA probe. Cells were incubated for 24 h with either drug-free medium or medium containing free MiR, free COD, a combination of MiR and COD, or PLGA@MiR&COD. After that, cells were incubated with DCFH-DA (10 µM) for 20 min and then analyzed using CLSM to capture fluorescence images.

Determination of intracellular drug concentration

SKOV3-TR cells were plated in 6-well plates. After 24 h of initial incubation, cells were exposed to drug-free medium, free MiR, MiR + COD, or PLGA@MiR&COD-containing medium, followed by an additional incubation period of 2–8 h. Upon completion of the co-incubation, the cells were digested, and 5 × 10⁵ cells were collected from each group. Subsequently, the cell samples were digested with H₂O₂, concentrated HCl, and concentrated nitric acid. Finally, the samples were diluted to a volume of 4 mL using diluted hydrochloric acid, and the platinum concentration in the cell samples was measured via the ICP-MS. Calibration was performed using a series of platinum standard solutions at concentrations of 0, 0.1, 0.2, 0.5, 1.0, and 10.0 µg/L. The calibration curve demonstrated excellent linearity, with a correlation coefficient exceeding 0.999. Matrix effects were effectively minimized through complete sample digestion and appropriate dilution. The detection limit for platinum was established at 0.01 µg/L. For the rhodamine 123 accumulation assay, SKOV3-TR cells were treated with free COD or PLGA@COD formulations for 24 h. After incubation, rhodamine 123 (5 µg/mL) was added and cells were incubated for an additional 30 min at 37°C. Cells were then washed with PBS, collected, and analyzed by flow cytometry to assess intracellular fluorescence intensity.

Mitochondrial membrane potential and ATP level determination

The changes in mitochondrial transmembrane potential induced by different treatments were assessed using the JC-1 probe. SKOV3-TR cells were treated with drug-free medium, free MiR, free COD, MiR + COD, or PLGA@MiR&COD-containing medium. The concentration of COD and MiR was set at 40 µg/mL. After a 24-hour incubation with the agents, cells were stained with JC-1 solution (2 µM) and subsequently washed twice with cold PBS to remove excess dye. The samples were subsequently analyzed using CLSM, with excitation at 488 nm and emission at 530 nm for green fluorescence, and excitation at 525 nm and emission at 590 nm for red fluorescence.

For intracellular ATP analysis, the SKOV3-TR cells were treated with drug-free medium, free MiR, free COD, MiR + COD, or PLGA@MiR&COD-containing medium. After co-incubation of the cells with the drugs for 12–24 h, a luminescent ATP detection assay kit was used to analyze the intracellular ATP level.

Apoptosis analysis

Apoptosis in SKOV3-TR cells was assessed using the FITC Annexin V Apoptosis Detection Kit I. SKOV3-TR cells were treated with either drug-free medium, free MiR, free COD, MiR + COD, or a medium containing PLGA@MiR&COD for 24 h. Both COD and MiR were set at a concentration of 40 µg/mL. All subsequent procedures were performed according to the reagent instructions. Cell apoptosis was then analyzed by flow cytometry.

Wound healing and transwell migration assay

To synchronize the cells and reduce proliferation, SKOV3-TR cells were serum-starved in medium containing 0.5% FBS for 24 h. A linear scratch was then made across the cell monolayer using sterile 10 µL pipette tips. After washing with PBS to remove detached cells, the cultures were incubated in complete medium with (free MiR, free COD, MiR + COD, or PLGA@MiR&COD) or without drugs at 37°C in a 5% CO₂ incubator for 3 to 12 h. Wound closure was monitored using the brightfield mode of an inverted fluorescence microscope (Leica, Germany).

To assess migration, SKOV3-TR cells were seeded into transwell inserts and incubated with 10% FBS in the lower chamber as a chemoattractant. After 24–48 h, migrated cells were fixed, stained, and counted under a microscope.

Establishment of drug-resistant ovarian cancer peritoneal metastasis models

A platinum-resistant ovarian cancer metastasis model was established by intraperitoneal injection of 2.0 × 10⁶ SKOV3-TR-luc cells into 5-week-old female BALB/c nude mice. Mice were housed under SPF conditions, and the model was used 12 days post-injection once the abdominal fluorescence intensity reached 1.0 × 10⁸ p/s/cm²/sr.

In vivo pharmacodynamics study

Mice were randomly divided into six groups (n = 6) and treated every three days with saline, free MiR, free COD, free MiR + COD, PLGA@MiR, or PLGA@MiR&COD. Antitumor efficacy was monitored by bioluminescence imaging on days 12, 20, 24, 27, and 31. On day 32, blood samples were collected for hematological and biochemical analysis, and metastatic sites in the abdominal cavity were examined and measured. For survival analysis, xenograft models were established with 4.0 × 10⁶ SKOV3-TR-luc cells per mouse, and treated as above. Kaplan–Meier analysis was used to evaluate survival outcomes.

Determination of tumor tissue cholesterol and ROS levels

The tumor tissue obtained from the animal experiment in each group was fixed with formalin and paraffin. Subsequently, the tissue samples were stained with DCFH-DA dye or Filipin fluorescent dye. Dewaxing and sample staining were performed according to the reagent instructions. The slides were then scanned and analyzed using a fluorescence slide scanner (Olympus, Japan).

Quantitative RT-PCR for p-gp mRNA expression in tumor tissues

To assess p-gp mRNA expression, one peritoneal metastatic tumor nodule was collected from each nude mouse, immediately snap-frozen in liquid nitrogen, and homogenized in TRIzol reagent. Total RNA was then extracted, and cDNA was synthesized using a reverse transcription kit. Quantitative real-time PCR was performed using a Bio-Rad PCR system (California, USA). The mRNA expression of p-gp was measured using specific primers (forward: 5’-TCTGGCATTCTGGTATGGGACT-3’; reverse: 5’-CTGGTTTGTGCCCACTCTTTG-3’), with GAPDH as the internal control (forward: 5’-CCTCGTCCCGTAGACAAAATG-3’; reverse: 5’-TGAGGTCAATGAAGGGGTCGT-3’).

Immunohistochemical analysis of tumor tissues

Tumor tissues were fixed, paraffin-embedded, and sectioned for H&E, TUNEL, and Ki67 staining. H&E staining followed standard procedures. TUNEL staining was performed using a commercial kit, with apoptotic cells visualized as red fluorescence after DAPI counterstaining. Ki67 immunohistochemistry involved antigen retrieval, blocking, incubation with primary antibody, and detection using the 3,3’-diaminobenzidine substrate kit, followed by hematoxylin counterstaining. All sections were imaged using a digital scanning microscope (Grundium, Finland).

Serum toxicity evaluation

To assess the serum toxicity of treatments, plasma samples from mice in all groups were analyzed. The concentrations of alanine aminotransferase (ALT), aspartate aminotransferase (AST), and blood urea nitrogen (BUN) in the plasma were determined using a fully automated blood biochemical analyzer (Hitachi, Japan).

Evaluation of major organ toxicity

To assess normal tissue damage, the major organs (heart, liver, spleen, lung, and kidney) from each group in the pharmacodynamics analysis were harvested and weighed. The tissue was fixed in 4% paraformaldehyde for 24 h, followed by sectioning and histological analysis using H&E and Masson’s trichrome staining to assess tissue morphology and fibrosis.

Statistical analysis and mapping software

The normality of continuous variables was assessed using the Kolmogorov–Smirnov test. For comparisons between two groups, Student’s t-test was applied to normally distributed data, while the Mann–Whitney U test was used for non-normally distributed data. Categorical variables were compared using the Chi-square test. For pairwise comparisons following Chi-square analysis, the Chi-square partition method was applied. When expected frequencies in contingency tables were less than 5, Fisher’s exact test was used; otherwise, the Chi-square test was performed.

To evaluate the diagnostic or predictive performance of significant variables identified in previous comparative analyses, receiver operating characteristic (ROC) curves were Generated. A variable was considered to have discriminative potential when the 95% confidence interval (CI) of the area under the curve (AUC) did not include 0.5.

Trajectory analysis and visualization of repeatedly measured continuous data were conducted using Stata software (version 18.0; StataCorp LLC, College Station, TX, USA). Group comparisons involving more than two groups were performed using one-way analysis of variance with SPSS software (version 25.0; IBM Corp., Armonk, NY, USA). Continuous data are presented as mean ± standard deviation (SD). A two-tailed p-value < 0.05 was considered statistically significant, while p-value < 0.01 indicated a highly significant difference.

Results and discussion

Clinical description of EOC patients

Existing reports have documented elevated cholesterol levels in breast cancer, bladder cancer, and colorectal cancer [33–35]. Nevertheless, cholesterol levels should be validated using clinical data and patient-derived samples before applying cholesterol-depleting strategies. In this study, 192 patients with EOC were enrolled. Among them, 48 were classified as platinum-resistant, while 144 were identified as platinum-sensitive [29]. All patients had undergone primary debulking surgery and completed the first course of platinum chemotherapy in the Women’s Hospital, Zhejiang University School of Medicine (Hangzhou, China). Among the 192 patients, the number of available pre-chemotherapy serum cholesterol measurements ranged from 1 to 29, with a median value of 6. The patient with the most cholesterol measurements had 29 assessments taken at various time points and was diagnosed with EOC at the age of 51. Recurrence occurred 28 months after completing the last platinum-based chemotherapy.

Serum cholesterol detection in EOC patients

Among the 192 patients, a total of 172 underwent preoperative serum cholesterol evaluation. The serum cholesterol levels were measured in 101, 138, 134, 131, 127, and 116 patients before the first to sixth cycles of chemotherapy, respectively. Serum cholesterol levels decreased in both drug-sensitive and drug-resistant groups following the first chemotherapy cycle (Fig. 1A, B and Fig. S1). Thereafter, the cholesterol levels of platinum-sensitive EOC patients increased and then slightly decreased, showing a fluctuating pattern (Fig. 1A). In contrast, the cholesterol levels of platinum-resistant EOC patients consistently increased starting from the second chemotherapy cycle (Fig. 1B). The preoperative serum cholesterol levels were found to be essentially similar between the two groups (Fig. 1C). However, the preoperative serum cholesterol level in the resistant group (R) before the sixth cycle (5.66 ± 1.15 mmol/L) was higher than that in the sensitive group (CR) (5.25 ± 1.02 mmol/L), although this difference was not statistically significant (p = 0.078) (Fig. 1D). When patients in both groups were stratified by age into ≤ 50 and > 50 years old, the cholesterol levels of the sensitive group in both age categories displayed a fluctuating pattern, while the cholesterol levels of the resistant patients showed an increasing trend from the first chemotherapy cycle (Fig. S2-3). Following the sixth cycle of chemotherapy, serum cholesterol levels in the resistant group were higher compared to the sensitive group (p = 0.130 for ≤ 50 years old group and p = 0.205 > 50 years old group). Overall, the serum cholesterol levels of patients with platinum-resistant EOC increased with the progression of chemotherapy cycles from the first to the sixth cycle.

Fig. 1.

Serum cholesterol levels in EOC patients with complete response (A) and 19 chemo-resistant EOC patients (B) measured before surgery and prior to each of the first to sixth chemotherapy cycles. (C) Preoperative serum cholesterol levels in platinum-sensitive (labeled as CR) and platinum-resistant (labeled as R) EOC patients. (D) Serum cholesterol levels prior to the sixth cycle of chemotherapy in platinum-sensitive and platinum-resistant EOC patients. Filipin staining of tumor tissues from platinum-sensitive (E) and platinum-resistant (F) EOC patients. (G) Fluorescence intensity distribution of Filipin-stained cholesterol in ovarian cancer tissues from both groups. Trajectory analysis of serum cholesterol levels during chemotherapy in platinum-resistant (H) and platinum-sensitive (I) patients

Cholesterol measurement in EOC tumor tissues

The cholesterol measurement assay included 15 patients, comprising 8 with platinum-resistant tumors and 7 with platinum-sensitive tumors. After exclusion, no significant differences were observed between the two groups in terms of FIGO stage, age, and tumor grade (Table S1). Filipin staining was employed to assess cholesterol levels in tumor tissues from both patient groups (Fig. 1E-G). The findings demonstrated that the cholesterol content in ovarian tumor tissues of the platinum-resistant group was significantly higher compared to the platinum-sensitive group (p < 0.05). In comparison to serum findings, cholesterol levels were more profoundly elevated in platinum-resistant ovarian tumor tissues compared with platinum-sensitive ones.

Analysis of serum cholesterol levels predicts platinum resistance

To reveal the changes in serum cholesterol levels during chemotherapy and their relationship with platinum sensitivity, this study conducted a trajectory analysis of serum cholesterol levels in patients with platinum-resistant and platinum-sensitive tumors. The computer simulation divided the patients in both groups into three categories based on their cholesterol levels: high (green line), medium (red line), and low (blue line). The findings demonstrated significant differences in cholesterol level changes between patients with platinum-sensitive tumors and those with platinum-resistant tumors (Fig. 1H, I). In the low cholesterol level group (blue line), the trajectory of patients sensitive to chemotherapy showed an upward trend in the early stages of treatment (approximately the first 10 time points), then tended to stabilize or slightly decrease. Conversely, the trajectory of drug-resistant patients rapidly increased at the beginning of treatment, reaching a peak before starting to decline, with great fluctuations throughout the treatment process. Additionally, a higher proportion of patients with low serum cholesterol levels was observed in the chemotherapy-sensitive group (24.0% for green line), while the proportion of patients with low cholesterol levels in the drug-resistant group was significantly lower, at only 6.3%. The proportions of patients with high serum cholesterol levels were comparable between the two groups, with the main difference observed in the population with medium cholesterol levels (55.5% in the drug-resistant group and 38.2% in the sensitive group). These differences highlighted the abnormality of cholesterol metabolism in drug-resistant patients during treatment, suggesting that serum cholesterol levels could serve as an important indicator for evaluating drug resistance.

Based on the cholesterol staining results of EOC tumors, we could also predict the resistance of EOC patients to platinum drugs using the ROC curve of Filipin fluorescence density (Fig. S4). When the cut-off value was set at 2.022, the Youden index reached its peak (marked by the arrow), exhibiting a sensitivity of 75.00% and specificity of 85.70%. Consequently, patients with tissue Filipin fluorescence density greater than 2.022 are more likely to develop platinum resistance. With an increasing number of patients, the two predictive models will emerge as valuable tools for anticipating drug resistance and assessing clinical outcomes, thus informing and optimizing treatment strategies. The above results demonstrated that cholesterol levels in ovarian cancer tissues of platinum-resistant patients were higher than those of platinum-sensitive patients. Therefore, regulating the level of plasma membrane cholesterol may represent a potential strategy for managing advanced ovarian cancer with drug-resistant features.

Preparation and characterization of drug-loaded PLGA microspheres

While cisplatin and carboplatin are standard platinum-based agents for ovarian cancer, their use in recurrent or platinum-resistant cases is often limited by cumulative toxicities such as nephrotoxicity and gastrointestinal side effects. Miriplatin, a lipophilic second-generation platinum compound, was developed to reduce these adverse effects and has been approved in several countries, including Japan, as an alternative to conventional platinum drugs [36–38]. Moreover, locoregional delivery of cisplatin or carboplatin can cause significant off-target toxicity. Therefore, miriplatin was chosen for this study. Accordingly, PLGA microspheres co-loaded with COD and MiR (PLGA@MiR&COD) were engineered (Fig. 2A) to evaluate their therapeutic efficacy and underlying mechanisms in advanced ovarian cancer animal models (Fig. 2B-E). The SEM image showed that both the blank and drug-loaded PLGA microspheres were quasi-spherical (Fig. 3A, B). The unloaded PLGA microspheres exhibited a mesoporous structure with high porosity and pore sizes ranging from 100 to 300 nm. After loading with MiR and COD, the SEM and TEM images showed that the pores on the surface of the PLGA microspheres vanished, attributed to the drugs filling the pores of the PLGA microspheres (Fig. 3B, C). The dispersion of PLGA@MiR&COD displayed a blue opalescence and a significant Tyndall effect (insert images in Fig. 3C). The blank PLGA microspheres exhibited a mean particle size of approximately 2280 nm, a polydispersity index (PDI) of 0.05, and a zeta potential of around − 0.57 mV. The incorporation of MiR and COD resulted in a slight increase in the particle size of the PLGA microspheres, reaching approximately 2580 nm. PLGA@MiR&COD exhibited a PDI of 0.06 and a zeta potential of approximately − 0.78 mV (Fig. 3D-G). Stability studies showed that the particle size of PLGA@MiR&COD exhibited minimal changes over 7 days in the environment at pH 7.4 with or without 10% FBS (Fig. S5), indicating good colloidal stability under physiological conditions. Encapsulation of MiR and COD within PLGA microspheres enhanced safety by enabling controlled release and avoiding the sharp post-administration spike in drug concentration that could damage normal tissues. In the hemolysis assay, PLGA@MiR&COD at a concentration of 100 µg/mL showed a hemolysis rate below 0.5%, demonstrating excellent biocompatibility (Fig. S6).

Fig. 2.

Schematic illustration of PLGA@MiR&COD preparation (A), intraperitoneal administration (B), ovarian cancer peritoneal metastasis (C), and the proposed mechanism of cholesterol oxidation-mediated membrane lipid therapy (D, E)

Fig. 3.

(A) Representative SEM image of blank PLGA microspheres. Representative SEM (B) and TEM (C) image of PLGA@MiR&COD. Insert: morphological changes of PLGA@MiR&COD during preparation via the dual emulsion method. From left to right: crude emulsion, microsphere suspension, and Tyndall effect observed under laser irradiation. Representative histogram of particle size distribution and zeta potential of blank PLGA microspheres (D, E) and PLGA@MiR&COD (F, G). (H) SDS-PAGE analysis of cholesterol oxidase encapsulated in PLGA@MiR&COD. Lanes 1–4 refer to PBS, free COD, PLGA@MiR&COD, and washed PLGA@MiR&COD. (I) Cumulative release profiles of MiR from PLGA@COD&MiR under different conditions (n = 3)

Encapsulation efficiency determination and in vitro release study

PLGA@MiR&COD achieved an encapsulation efficiency of 82.98 ± 0.09% for MiR and 32.83 ± 4.30% for COD. In the PLGA@MiR&COD formulation, the mass ratio of MiR to COD was 1.01. Sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) analysis revealed that, under equal COD loading conditions, the intensity of the protein band (55 kDa) in the freshly prepared PLGA@MiR&COD group (lane 3) was markedly lower than that in the free COD group (lane 2), indicating successful encapsulation of COD into PLGA microspheres (Fig. 3H). Notably, after thorough washing and overnight storage, no visible COD band was observed in the PLGA@MiR&COD group (lane 4), suggesting that unencapsulated COD was effectively removed and the encapsulated enzyme was stably retained within the microspheres without obvious leakage in aqueous conditions.

The in vitro release study demonstrated that PLGA@MiR&COD exhibited sustained drug release behavior, with cumulative release rates at 48 h reaching 57.1 ± 7.7% in pH 7.4 without FBS, 51.6 ± 5.4% in pH 7.4 with 10% FBS, and 63.5 ± 9.8% in pH 6.5 with 10% FBS (Fig. 3I). The accelerated drug release observed under mildly acidic conditions (pH 6.5), which mimic the tumor microenvironment, was likely due to the faster degradation of PLGA in acidic environments. Moreover, the presence of 10% FBS slightly reduced the release rate at pH 7.4 environment, possibly due to the formation of a protein corona on the microsphere surface, which hindered PLGA degradation and delayed diffusion.

Cellular uptake and intracellular localization

Figure 4 A and B displayed that the cellular uptake of free Cy5 and PLGA@Cy5 by SKOV3-TR cells was time-dependent, ranging from 0.25 to 8 h, with the uptake gradually increasing as time progresses. At all examined time points, SKOV3-TR cells exhibited higher uptake of free Cy5 compared to PLGA@Cy5, which may be attributed to the small molecular size and lipophilic nature of Cy5 that facilitate its passive diffusion into cells. The confocal results were consistent with the flow cytometry studies, showing a gradual increase in the uptake of PLGA@Cy5 by SKOV3-TR cells over time (Fig. 4C). Furthermore, PLGA@Cy5 was mainly distributed in the cytoplasm, with almost no red fluorescence detected in the cell nucleus.

Fig. 4.

(A) Flow cytometry of SKOV3-TR cells treated with free Cy5 or PLGA@Cy5. (B) Quantification of mean fluorescence intensity (n = 3). (C) CLSM images of SKOV3-TR cells after different treatments. Cell viability after 24-hour (D) or 48-hour (E) treatment with free MiR, free COD, MiR + COD, or PLGA@MiR&COD. (F) IC₅₀ comparison among treatment groups. **p < 0.01, ####p < 0.0001 vs. free MiR group, ★★★★p < 0.0001 vs. free COD group, ▲▲▲▲p < 0.0001 vs. MiR + COD group

Micron-sized PLGA particles are primarily internalized via endocytic pathways and subsequently trafficked to endo-lysosomal compartments [39]. Within these acidic environments, the PLGA matrix undergoes gradual hydrolytic degradation, enabling the controlled release of COD while protecting it from premature enzymatic inactivation. COD remains enzymatically active at mildly acidic pH (~ 5.0–6.0) [40], allowing it to oxidize cholesterol within endosomal and lysosomal membranes. This oxidative process compromises membrane integrity, promotes the leakage of lysosomal enzymes [41], and disrupts endocytic recycling, thereby facilitating the cytosolic release of both COD and MiR. These combined effects may contribute to enhanced chemo-sensitization.

Cellular toxicity analysis

As displayed in Fig. 4D-F, the half-maximal inhibitory concentrations (IC₅₀) of MiR and COD were determined to be 81.19 ± 7.16 µg/mL and 216.88 ± 19.00 µg/mL, respectively, following 24 h of co-culture with SKOV3-TR cells. In both the physical mixture (MiR + COD) and the PLGA@MiR&COD formulation, the mass ratio of MiR to COD was maintained at 1:1. Based on the MiR concentration, the IC₅₀ values for MiR + COD and PLGA@MiR&COD were calculated as 36.73 ± 2.25 µg/mL and 45.65 ± 8.44 µg/mL, respectively (Fig. 4D, F). These findings demonstrated that the presence of COD significantly enhanced the cytotoxic effect of MiR against SKOV3-TR cells. Similar conclusions could be drawn from the 48-hour cell toxicity analysis (Fig. 4E, F). The above analysis revealed that the cellular toxicity of PLGA@MiR&COD was lower than that of the physical mixture formulation of MiR and COD, which could be attributed to two factors: (i) MiR and COD directly interacted with cells and could enter SKOV3-TR cells by diffusion, while PLGA@MiR&COD needed to be internalized into cells before it exerted effect; (ii) In vitro release analysis demonstrated that PLGA@MiR&COD exhibited a sustained-release profile, resulting in a slower drug release compared to free drug formulations.

Evaluation of cholesterol and lipid raft levels of plasma membrane

To elucidate the mechanism underlying the enhanced cytotoxicity of the formulations, Filipin staining was used to assess cellular cholesterol levels, where increased blue fluorescence indicated higher cholesterol content. The β-CD forms a well-characterized 1:1 inclusion complex with cholesterol under physiological conditions [42]. This strong and specific binding interaction has been extensively studied and serves as the basis for its widespread use as cholesterol depleting agent [32, 43]. As displayed in Fig. 5A, Filipin staining of cells in each treatment group was observed to outline the cellular contours, with relatively weaker blue fluorescence in the cytoplasm and nucleus, demonstrating that cholesterol was mainly distributed on the plasma membrane of SKOV3-TR cells. In contrast to the control group, treatment with free MiR led to minimal alterations in intracellular Filipin fluorescence. A significant reduction in Filipin fluorescence on the cell membrane was noted upon treatment with the three drug formulations containing COD: the free COD group, MiR + COD group, and PLGA@MiR&COD group. Quantitative analysis (Fig. S7) showed consistent results, with all three COD-containing groups significantly reducing membrane cholesterol levels compared to the control (p < 0.01), confirming effective cholesterol depletion by both free COD and formulated COD.

Fig. 5.

CLSM images of cholesterol (A) and lipid raft (B) expression on SKOV3-TR cell membranes after different treatments. (C) Membrane fluidity analysis (n = 4). Western blot of p-gp expression (D) and its quantification (E) normalized to β-actin (n = 3). (F) Intracellular platinum levels (n = 5). (G) CLSM images of intracellular ROS. (H) Mitochondrial membrane potential detected by JC-1 staining; red and green indicate JC-1 aggregates and monomers, respectively. *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001 vs. blank group, #p < 0.05, ##p < 0.01, ####p < 0.0001 vs. free MiR group, ▲p < 0.05, ▲▲▲p < 0.001 vs. MiR + COD group, ▼p < 0.05 vs. PLGA@MiR group

Cholesterol is a key element in the lipid rafts of cell membranes, playing an important role in maintaining their structural integrity and function [44, 45]. In this study, AF488-CTB fluorescent dye was applied to assess the status of lipid rafts in plasma membrane following drug treatment. As depicted in Fig. 5B, upon treatment of SKOV3-TR cells with COD, MiR + COD, or PLGA@MiR&COD, a noticeable reduction in the fluorescence intensity of the plasma membrane lipid rafts was observed, suggesting that the structural integrity of the lipid rafts was compromised following the oxidation of cholesterol by COD. By comparison, cells in the control and free MiR groups exhibited a higher abundance of lipid rafts.

Changes in membrane lipid composition can significantly impact cellular functions such as endocytosis and receptor expression. Cholesterol and sphingolipids are essential for the formation of lipid raft microdomains, which support the structural integrity and membrane localization of receptors like Toll-like receptor 4 and fibroblast growth factor receptor 2b [46, 47]. Depleting these lipids disrupts raft organization and impairs receptor-mediated endocytosis. Receptors such as macrophage-inducible C-type lectin and interferon gamma receptor 1 also rely on intact lipid rafts for efficient internalization, and cholesterol depletion can block these pathways [48, 49]. Notably, inhibition of receptor-mediated endocytosis has been linked to reduced pro-tumor signaling, immune evasion, and drug resistance.

Lipid rafts enriched in cholesterol also support receptor recycling, helping maintain surface expression of receptors such as brassinosteroid-insensitive 1 and low-density lipoprotein receptor-related protein 1 [47]. Moreover, the cholesterol-sphingolipid balance regulates membrane fluidity, affecting receptor mobility and activation, especially in G protein-coupled receptors and receptor tyrosine kinases [50]. The ratio of glycerophospholipids to cholesterol influences membrane order, thereby modulating the localization and activation of receptors like vascular endothelial growth factor receptor 2 and epidermal growth factor receptor, both of which are involved in drug resistance pathways [17, 51]. These findings underscore the therapeutic value of targeting membrane lipid composition to counteract drug resistance and limit tumor progression.

Plasma membrane fluidity evaluation

Reduced cholesterol levels and disrupted lipid rafts can alter the structural properties of the tumor cell membrane [6]. The findings demonstrated that SKOV3-TR cells treated with the three COD-based drug formulations exhibited a notable increase in cell membrane fluidity compared to both the control group and the group treated with free MiR (Fig. 5C). This improvement in fluidity was attributed to the oxidation of cholesterol by COD, which consequently led to a decrease in lipid raft levels.

Analysis of drug-resistant protein and intracellular drug concentration

Cholesterol modulates the rigidity and fluidity of lipid rafts, influencing the conformation and function of embedded proteins like p-gp, and facilitating the export of chemotherapeutic agents [52, 53]. Disrupting cholesterol-rich rafts with small-molecule agents has been shown to reverse drug resistance [54]. In this study, we examined the expression levels of p-gp in SKOV3-TR cells across different treatment groups at both the protein (Fig. 5D, E) and mRNA levels (Fig. S8). The COD-mediated oxidation of cholesterol disrupted the structure of lipid rafts, thereby affecting p-gp expression. Notably, PLGA@MiR&COD significantly reduced the mRNA and protein expression of p-gp in tumor cells. The intracellular drug concentration assay showed that, at equal dosages, both the MiR + COD and PLGA@MiR&COD formulations achieved significantly higher intracellular accumulation than free MiR after 1 and 8 h of co-incubation (p < 0.05) (Fig. 5F). Moreover, the rhodamine 123 accumulation assay showed that both PLGA@COD and free COD treatments significantly increased intracellular fluorescence compared to rhodamine 123 alone, with the free COD group showing the highest signal (Fig. S9), indicating reduced drug efflux activity. These findings further corroborated our hypothesis that cholesterol oxidation by COD in lipid rafts could mitigate the drug efflux effect of drug-resistant OC cells by modulating the expression and function of p-gp protein.

Intracellular ROS level detection

The H₂O₂ produced during the oxidation of cholesterol can generate ·OH with high oxidation activity under the catalysis of endogenous Fenton-like catalysts, such as Fe²⁺/Fe³⁺ [14]. The results demonstrated that, compared to the control group, the levels of ROS in SKOV3-TR cells were elevated following treatment with free COD, MiR + COD, or PLGA@MiR&COD (Fig. 5G).

Mitochondrial membrane potential and apoptosis analysis

Compared to the MiR group, the intracellular red fluorescence of J-aggregates in the MiR + COD group and PLGA@MiR&COD group was reduced (Fig. 5H), suggesting that the combination of COD and MiR induced a decrease in mitochondrial transmembrane potential and compromised mitochondrial structure. Mitochondrial dysfunction severely disrupts cellular energy metabolism [55]. As displayed in Fig. 6A, the ATP content in the free MiR group showed a slight decrease, which could be due to cell damage caused by the MiR. Compared to the free MiR group, the free COD group exhibited a significant reduction in ATP content, indicating that COD could inhibit cellular energy metabolism. Notably, the ATP depletion in the MiR + COD group and the PLGA@MiR&COD group was comparable, demonstrating their effective inhibition of ATP production. Mitochondria are the main sites of ATP biosynthesis, providing energy for the operation of drug-resistant proteins like p-gp. Thus, mitochondrial dysfunction would compromise the activity of drug-resistance proteins. The increased intracellular platinum concentration after treatment with MiR + COD or PLGA@MiR&COD could be attributed to multiple effects of COD-containing formulations, including the disruption of lipid rafts and the blocking of energy supply. Cholesterol oxidation-derived generation of ROS also induced the apoptosis of tumor cells. Flow cytometry analysis showed that the proportion of late apoptosis (Q3) induced by COD + MiR in OC cells was markedly higher than that in the MiR group (p < 0.05) (Fig. 6B). Although the apoptosis-inducing effect of PLGA@MiR&COD was less pronounced than that of MiR + COD, it remained more effective than either free MiR or free COD alone. The specific reasons have been discussed in the cytotoxicity experiment section.

Fig. 6.

Intracellular ATP levels (A) and apoptosis analysis (B) of SKOV3-TR cells after different treatments (n = 3). (C) Wound healing images at indicated time points. (D) Cell invasion after 24 and 48 h, assessed by crystal violet staining. *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001 vs. blank group, #p < 0.05, ##p < 0.01, ###p < 0.001, ####p < 0.0001 vs. free MiR group, ★p < 0.05, ★★p < 0.01 vs. free COD group, ▲p < 0.05 vs. MiR + COD group

The above cytological studies demonstrated how cholesterol oxidation-mediated membrane lipid therapy enhanced the sensitivity of platinum-resistant OC cells (as shown in Fig. 2B-E). Specifically, COD extracted cholesterol from the plasma membrane, resulting in the disruption of lipid raft integrity. The consequent reduction in membrane fluidity impaired the structural support for p-gp, thereby attenuating its expression and efflux function. The ·OH radicals produced through cholesterol oxidation affected the energy supply to the p-gp. Thus, the intracellular MiR concentration increased following treatment with MiR + COD or PLGA@MiR&COD. Moreover, the ·OH radicals generated from cholesterol oxidation synergistically promote apoptosis of SKOV3-TR cells with MiR.

In vitro wound healing and transwell experiments

To assess the impact of PLGA@MiR&COD on the metastasis and invasion of SKOV3-TR cells, wound healing and transwell assays were conducted. The results demonstrated that the cell migration rate in the MiR monotherapy group was similar to that in the control group (Fig. 6C). Compared to the free MiR group, COD inhibited the invasion of SKOV3-TR cells, possibly due to several reasons: (i) Tumor cell invasion requires the support of a plasma membrane with certain hardness, and COD-induced cholesterol oxidation alters the physical properties of membrane, reducing its supportive effect on the cells; (ii) Tumor cell metastasis necessitates energy support, and the ·OH generated during cholesterol oxidation can block mitochondria from producing ATP, thereby inhibiting tumor metastasis; (iii) ·OH radicals lead to apoptosis of tumor cells and suppress the proliferation of SKOV3-TR cells. Notably, the wound healing rate in the MiR + COD group was significantly reduced compared to the other three groups. The wound healing rate in the PLGA@MiR&COD group was lower than that in the free MiR and free COD groups but slightly higher than in the MiR + COD group. The transwell experiment results were consistent with the wound healing experiment findings. After 24–48 h of co-incubation, the MiR + COD group showed a significantly lower number of invaded cells compared to the other three groups, suggesting that SKOV3-TR cell invasion was most effectively inhibited in this group, with the PLGA@MiR&COD group showing similar effects (Fig. 6D). Overall, cytological experiments suggested that the plasma membrane cholesterol oxidation strategy could reverse platinum resistance in OC cells and inhibit their metastasis and invasion.

Pharmacodynamic analysis

Chemotherapy is typically administered intravenously or orally, but the blood-peritoneal barrier limits drug delivery to the peritoneal cavity [56]. Intraperitoneal chemotherapy offers a direct delivery route to the abdominal cavity. In this study, we adopted I.P. administration for three main reasons: (i) Advanced ovarian cancer often involves widespread abdominal metastasis; I.P. delivery increases local drug concentration and improves therapeutic outcomes; (ii) Local injection reduces systemic drug exposure, thereby lowering side effects of platinum drugs and minimizing COD-induced damage to blood vessels and normal tissues; (iii) PLGA microspheres, with their large surface area and mesoporous structure [57, 58], can adsorb onto the peritoneum and organs, reducing clearance and prolonging drug retention.

The experimental design is detailed in Fig. S10A. The primary objective of this study was to investigate the synergistic therapeutic effect of MiR and COD. To align with ethical principles and minimize animal usage, the COD-loaded PLGA formulation was not included in the experimental design. Before the establishment of the peritoneal metastasis model, the background fluorescence in nude mice was negligible (Fig. S10B). Between days 12 and 20, there was a substantial increase in fluorescence intensity across all groups (Fig. 7A, B). Notably, the fluorescence growth rate in the MiR + COD, PLGA@MiR, and PLGA@MiR&COD groups was slower, demonstrating significant hindrance to tumor tissue proliferation post-administration. During the treatment period from day 20 to day 24, the average fluorescence intensity in the saline and free COD groups continued to rise, accompanied by an expansion in abdominal fluorescence area, highlighting the vigorous growth and increasing metastasis of drug-resistant OC cells within the abdominal cavity. In contrast, the fluorescence intensity in the free MiR and PLGA@MiR groups increased more slowly than the free COD group, demonstrating their tumor-inhibitory effects. Remarkably, a decline in fluorescence intensity was observed in the MiR + COD and PLGA@MiR&COD groups starting from day 20, suggesting that the combination therapy effectively eliminated drug-resistant ovarian cancer cells within the peritoneal cavity. By day 31, the fluorescence intensity of the PLGA@MiR&COD group was the lowest during the last fluorescence intensity monitoring. By day 45, some animals in the free MiR and MiR + COD groups exhibited signs of weakness and impaired feeding and drinking behaviors. In compliance with ethical regulations, all animal experiments were terminated, and the mice were humanely euthanized according to institutional guidelines. Relevant data were subsequently collected. Figure 7C showed the peritoneal metastatic nodules isolated from each mouse. The results demonstrated that most mice developed multiple ovarian cancer metastasis of varying sizes within the peritoneal cavity, with up to 27 nodules observed in one case. Notably, in the PLGA@MiR and PLGA@MiR&COD groups, one mouse in each group showed no visible peritoneal metastasis. Given that all included mice exhibited peritoneal fluorescence intensities exceeding 1.0 × 10⁸ p/s/cm²/sr, the absence of detectable nodules in these cases was likely attributable to the therapeutic effect of the administered formulations. The average number of metastatic lesions were 18.67 ± 5.79, 13.17 ± 5.34, 14.00 ± 6.69, 9.00 ± 5.18, 11.33 ± 8.48, and 4.33 ± 5.96 for saline, free MiR, free COD, MiR + COD, PLGA@MiR, and PLGA@MiR&COD groups, respectively (Fig. 7D). The mean mass of metastasis was 1.07 ± 0.28 g, 0.58 ± 0.28 g, 0.53 ± 0.21 g, 0.31 ± 0.16 g, 0.48 ± 0.34 g, and 0.13 ± 0.30 g, for saline, free MiR, free COD, MiR + COD, PLGA@MiR, and PLGA@MiR&COD groups, respectively (Fig. 7E). The trends in mean metastasis volume were consistent with the trends in mean metastasis weight in each group of model animals (Fig. 7F). The tumor mass and volume of each model animal were displayed in Fig. S11-12.

Fig. 7.

Bioluminescence images (A) and individual bioluminescence changes (B) of mice during treatment (n = 6). Photos (C), number (D), total mass (E), and total volume (F) of tumor tissues dissected from mice in each group (n = 6). (G) Kaplan–Meier survival curves of platinum-resistant ovarian tumor-bearing mice after different treatments (n = 6). *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001 vs. saline group, #p < 0.05 vs. free MiR group

These findings demonstrated that PLGA@MiR&COD exhibited the most superior therapeutic effect among all treatment groups, showing its profound therapeutic effect on animals with drug-resistant OC metastasis. In vitro, the IC₅₀ of the MiR + COD group was higher than that of PLGA@MiR&COD. However, in vivo, free MiR and free COD were rapidly diluted and cleared by body fluids. Moreover, COD, as a biological enzyme, is susceptible to degradation by endogenous enzymes. In contrast, PLGA@MiR&COD not only showed strong organ affinity after I.P. administration, enabling sustained drug release, but also enhanced the enzymatic stability of COD through encapsulation, contributing to a more favorable therapeutic outcome compared to the free drug combination. In the 90-day survival study, the two-drug combination therapy significantly prolonged the survival time of mice compared to the monotherapy groups, with PLGA@MiR&COD showing the most notable benefit (Fig. 7G).

Cholesterol, ROS, and p-gp analysis in tumor tissues

Subsequently, this study investigated the levels of cholesterol and ROS in tumor tissues. Notably, cholesterol fluorescence in tumor tissues was lower in all COD-containing groups compared to the control and free MiR groups (Fig. 8A), with statistically significant differences observed upon quantitative analysis (Fig. S13, p < 0.05). Among all treatment groups, the PLGA@MiR&COD group exhibited the lowest cholesterol level in tumor tissues, indicating its superior efficacy in depleting membrane cholesterol in vivo. Cholesterol was oxidized by COD to generate ·OH. Consistent with in vitro results, ROS levels in tumor tissue significantly increased following treatment with the three COD-based formulations, with the highest ROS levels observed in the PLGA@MiR&COD-treated tumors (Fig. 8B). The clearance of plasma membrane cholesterol and·OH-mediated mitochondrial damage ultimately affected the expression of p-gp. Considering that only 5 animals in 2 groups developed detectable metastatic lesions, we selected 5 mice from each group for analysis of p-gp expression. As shown in Fig. S14, the ranking of p-gp mRNA expression in tumor tissues was as follows: free MiR > PLGA@MiR > free COD ≈ free MiR + COD > PLGA@MiR&COD.

Fig. 8.

Representative fluorescence staining images of cholesterol (A) and ROS (B) in tumor tissues. Representative TUNEL (C), H&E (D), and Ki67 (E) staining images of tumor tissues illustrating cell apoptosis, tissue necrosis, and cell proliferation, respectively. Photos (F), quality (G), and area (H) of diaphragm of platinum-resistant ovarian tumor-bearing mice after different treatments (n = 6). *p < 0.05, ***p < 0.001, ****p < 0.0001 vs. saline group, ##p < 0.01 vs. free MiR group

TUNEL, H&E, and Ki67 staining of tumor tissues

The overproduction of ROS can induce apoptosis in tumor cells. In the saline group, only a few OC cells within the tumor tissue showed signs of apoptosis (Fig. 8C). In contrast, increased apoptotic signals were observed in OC tissues treated with free MiR, with even stronger fluorescence detected in the free COD treatment group. Notably, tissues treated with MiR + COD and PLGA@MiR exhibited significantly larger apoptotic areas compared to the free COD group. Among all treatment groups, PLGA@MiR&COD induced the most extensive apoptosis within the tumor tissues. In the H&E-stained sections, tumor cells in the saline group exhibited intact morphology and tight arrangement. In contrast, all drug-treated groups showed varying degrees of tumor cell shrinkage and looser organization. Among them, the PLGA@MiR&COD-treated group displayed the most severe cellular damage (Fig. 8D).

Ki67 immunohistochemical staining showed a reduced number of Ki67-positive cells in the MiR + COD, PLGA@MiR, and PLGA@MiR&COD treatment groups than the saline and single drug groups, indicating decreased proliferative capacity of tumor cells. Notably, the PLGA@MiR&COD group exhibited the lowest Ki67 positivity rate (Fig. 8E). In clinical ovarian cancer samples, the Ki67 positivity rate serves as a predictive marker of platinum resistance, with lower Ki67 expression correlating with reduced resistance [59]. In this study, the decreased Ki67 positivity following PLGA@MiR&COD treatment suggested that this formulation could enhance tumor sensitivity to platinum-based chemotherapy.

Effects of plga@mir&cod on septal tumor colonization

A notable phenomenon observed in this study was the unique colonization pattern of SKOV3-TR cells in the diaphragm. Compared to the saline group, the area and mass of OC colonization in the diaphragm decreased in all drug treatment groups, with the PLGA@MiR&COD treatment group exhibiting the smallest membrane mass and overall mass (Fig. 8F-H). Figure S15 illustrated the TUNEL staining of diaphragm tissue. Areas with dense blue fluorescence indicated nuclei and represent regions of tumor colonization on the diaphragm. Areas with sparse blue fluorescence corresponded to normal diaphragm tissue. Red fluorescence signified apoptotic cells. Among the six groups, the MiR + COD group showed a smaller area of blue fluorescence accumulation compared to the saline, free MiR, and free COD groups. The MiR + COD-treated tumor exhibited stronger apoptosis signal than both single drug groups, showing that MiR + COD effectively induced apoptosis in tumor cells and reduced tumor colonization in the diaphragm. The PLGA@MiR&COD group had almost no blue fluorescence accumulation, suggesting the strongest inhibition of tumor cell colonization in the diaphragm. Ki67 and H&E staining of the diaphragm further confirmed the inhibitory effect of PLGA@MiR&COD on tumor cell colonization in the diaphragmatic tissues (Fig. S16-17). In general, MiR + COD, PLGA@MiR, and PLGA@MiR&COD effectively inhibited the colonization of drug-resistant OC cells in diaphragm tissue, with PLGA@MiR&COD demonstrating the most pronounced inhibitory effect.

Serological biosafety assessment

Nephrotoxicity, cardiotoxicity, neurotoxicity, and ototoxicity are common adverse reactions associated with platinum-based chemotherapy for OC, which seriously affect the life quality of patients [60, 61]. For example, kidney injury can progress to renal failure, requiring suspension of treatment or even renal replacement surgery [62]. Therefore, the safety of platinum-based drugs cannot be ignored, and the development of drug formulations must balance efficacy with potential risks to minimize platinum-related adverse effects and ensure safer patient care. This study assessed the serological toxicity of I.P. administration of PLGA@MiR&COD. Serum levels of alanine aminotransferase (ALT), aspartate aminotransferase (AST), and blood urea nitrogen (BUN) showed no significant differences across the six groups (Fig. 9A-C). However, multiple administrations of free MiR and MiR + COD led to higher mean values for these three indices compared to the two PLGA groups, especially with regard to ALT levels. Throughout the weight monitoring period, starting from day 20, the body weight curves of the six groups diverged into two distinct patterns: (i) The saline group and the two PLGA groups exhibited a gradual increase in animal body weight, which then stabilized; (ii) The free MiR, free COD, and MiR + COD groups experienced a decrease in body weight (Fig. 9D). Additionally, during the study, the mice in the MiR and MiR + COD groups displayed reduced activity and lethargy since day 30, which we attributed to the side effects of MiR.

Fig. 9.

Serum ALT (A), AST (B), and BUN (C) levels in mice (n = 6). (D) Body weight changes during treatment (n = 6). Organ weights of heart (E), liver (F), spleen (G), lung (H), and kidney (I) after treatment (n = 6). (J) H&E staining of major organs; yellow, blue, green, and red arrows indicate fibrosis, edema, inflammatory infiltration, and necrosis, respectively. *p < 0.05 vs. saline group, ▼▼p < 0.01 vs. PLGA@MiR group

Evaluation of normal tissue damage

The body weight and the mass of key organs, such as the heart, liver, spleen, lung, and kidney, were evaluated to determine the toxicity effects of the treatments (Fig. 9E-I). The results demonstrated that all administration groups did not exhibit significant changes in tissue weight of the liver, lung, and kidney of the mice. However, a notable decrease in heart mass was observed in the free MiR group, and the average spleen mass increased in the MiR + COD group. Spleen hypertrophy was observed in the MiR + COD group (Fig. S18). It was possible that the toxicity of MiR + COD might increase spleen permeability, thereby inducing edema.

Subsequently, the primary organs from mice in each group were harvested and subjected to H&E staining to assess potential organ toxicity (Fig. 9J). The staining revealed significant tissue damage in the free MiR and MiR + COD groups. After multiple administrations of MiR, signs of cardiac tissue fibrosis, interstitial thickening, septal edema, loss of cellular structure in certain areas, and severe tissue necrosis were observed. Additionally, there was massive swelling of liver cells. Lung tissue exhibited fibrosis, inflammation, and necrosis, with aggregated inflammatory cells and blurred structure in some areas. The kidney showed partial collapse with a small amount of glomerulus.

After treated with MiR + COD, the local interstitial thickened and fibrosis appeared. Many cardiomyocytes lost their normal morphology, resulting in heart tissue necrosis (Fig. 9J). In the liver, an increase in small round cells signified an ongoing inflammatory response. Pulmonary interstitial thickening and slight blurring of tissue structure demonstrated the presence of fibrosis and necrosis in the lungs. In the kidneys, glomerular edema and a noticeable aggregation of inflammatory cells were observed. The free COD group exhibited a significant accumulation of inflammatory cells in the lungs. Conversely, both PLGA groups displayed only mild inflammation in renal tissue. In this study, cholesterol oxidation by COD was passively regulated via the sustained-release effect of PLGA microspheres. However, this strategy did not enable precise spatial or temporal control over COD activity, and consequently, normal cells may also be affected. The serological and histopathological analyses revealed that all three COD-containing formulations, including the PLGA-encapsulated group, induced varying degrees of damage to normal tissues. This highlighted the need to enhance tumor-specific delivery in future studies. One potential approach is to modify the carrier with tumor-targeting ligands, such as human epidermal growth factor receptor 2 antibodies, or to design drug-releasing systems responsive to the ovarian tumor microenvironment.

Normal tissue fibrosis toxicity assessment

Masson staining was conducted on heart and spleen tissue sections to assess the fibrotic side effects induced by the drugs. As depicted in Fig. S19, the heart tissue of the MiR + COD group showed serious fibrosis, while the free MiR group and PLGA@MiR group displayed mild fibrosis. In addition, the MiR + COD group showed pronounced myocardial tissue edema, with a blurred interface and more severe inflammatory infiltration. In contrast, no evident myocardial tissue fibrosis was observed in the saline, free COD, and PLGA@MiR&COD groups. In the Masson-stained sections of spleen tissue, the fibrosis induced by different treatments was not obvious, with the free MiR and PLGA@MiR groups showing only mild spleen fibrosis (Fig. S20).

The safety evaluation results demonstrated that the MiR and MiR + COD groups experienced significant damage to normal tissue. In contrast, encapsulating MiR within PLGA effectively reduced the exposure of normal tissue to high concentration of drugs, thereby minimizing toxic side effects. The study presented in this section demonstrated that PLGA@MiR&COD could mitigate the toxicity and side effects associated with MiR significantly. Nevertheless, the safety of all drug formulations must be thoroughly validated before clinical translation.

Conclusion

Retrospective analysis of EOC patients revealed a progressive increase in serum cholesterol levels during the first six chemotherapy cycles, with higher tissue cholesterol levels observed in platinum-resistant tumor tissues compared to platinum-sensitive ones. These findings highlighted the potential of targeting plasma membrane cholesterol as a therapeutic strategy for advanced ovarian cancer. The designed PLGA@MiR&COD restored the sensitivity of drug-resistant ovarian cancer cells to MiR and inhibited metastatic colonization by inducing cholesterol oxidation in the plasma membrane. Therefore, PLGA@MiR&COD exhibited satisfactory therapeutic effects on advanced ovarian cancer. Moreover, the preparation reduced MiR-associated adverse effects. In summary, this study proposed a clinically relevant and innovative therapeutic strategy based on cholesterol-targeted membrane lipid therapy, laying a promising foundation for future translational research and personalized treatment of advanced ovarian cancer.

Author contributions

W.F. and Y.X. contributed equally to this work. W.F.: Writing – original draft, Investigation, Funding acquisition. Y.X.: Methodology, Investigation. W.Q.: Methodology, Investigation. M.L.: Formal analysis, Investigation, Funding acquisition. C.Z.: Investigation, Methodology. Y.L.: Investigation. D.C.: Investigation. Y.Z.: Investigation, Funding acquisition. S.X.: Validation, Methodology, Funding acquisition. X.W.: Writing – review & editing, Supervision, Conceptualization, Funding acquisition. M.Z.: Writing – review & editing, Supervision, Project administration, Funding acquisition, Conceptualization. All authors have discussed the results and approved the final manuscript.

Funding

This study was financially supported by the National Natural Science Foundation of China (Grant No. 82103505, 82071616, and 82002728), the Natural Science Foundation of Zhejiang Province (Grant No. ZCLY24H1602 and LQ23H180006), the Medical and Health Research Project of Zhejiang Province (Grant No. 2024KY192), and the Open Fund from Zhejiang Key Laboratory of Precision Diagnosis and Therapy for Major Gynecological Diseases (Grant No. ZDFY2024-MG-2).

Data availability

No datasets were generated or analysed during the current study.

Declarations

Ethics approval and consent to participate

This study received approval from the Ethics Committee of the Women’s Hospital, Zhejiang University School of Medicine (Ethical batch number: IRB-20200227-R). All animal experiment protocols were approved by the Scientific Investigation Board of Zhejiang Chinese Medical University (Ethics Number IACUC-20211122-05).

Consent for publication

All authors of this study agree to publish.

Competing interests

The authors declare no competing interests.