Therapeutic translation of traditional Chinese medicine Huangqi derived exosome like nanoparticles: targeting prostate cancer through ferroptosis activation, immune reprogramming, and microbiome modulation

Abstract

Plant derived exosome-like nanoparticles (PELNs) are emerging as a powerful tool for treating cancers. Among them, PELNs derived from traditional Chinese medicines have shown great potential in treating various cancers. However, many Chinese medicines remain to be explored, and currently, there are no reports on the use of PELNs for treating prostate cancer. In this study, we extracted Huangqi derived exosome-like nanoparticles (HELNs) from fresh Huangqi (Astragalus membranaceus) and systematically explored the effects and mechanisms of HELNs in treating prostate cancer through an integrated approach of single-cell sequencing, 16 S rDNA sequencing, and bulk-RNA sequencing. We found that HELNs demonstrated robust cytotoxic effects against prostate cancer both in vitro and in vivo. HELNs reverse the polarization of M2 macrophages, promote their M1-like polarization, and increase the abundance of anti-tumor probiotics in the gut to exert anti-tumor effects. In terms of direct effects, HELNs downregulate the expression level of GPX4 protein in prostate cancer cells, thereby inducing ferroptosis. Finally, we loaded siGPX4 into HELNs to construct a new nanoparticle siGPX4@HELNs. siGPX4@HELNs induce more pronounced ferroptosis in prostate cancer cells, achieving better therapeutic outcomes while maintaining safety.

Graphical abstract

Supplementary Information

The online version contains supplementary material available at 10.1186/s12951-026-04160-4.

Introduction

Prostate cancer (PCa) is one of the most common malignant tumors in the male urinary and reproductive system worldwide, ranking second in incidence and fifth in mortality among all male cancers. Global estimates for 2020 indicated approximately 1.4 million new cases and 370,000 deaths from prostate cancer [1]. In China, driven by an aging population and lifestyle changes, the incidence and mortality of prostate cancer continue to rise, posing a significant threat to the health of Chinese men [2]. According to the latest data from the National Cancer Center, China reported about 134,000 new cases and 47,000 deaths from prostate cancer in 2022 [3]. Depending on the stage of prostate cancer progression, various treatment options are available, including radical prostatectomy, endocrine therapy, radiotherapy, and chemotherapy. Endocrine therapy represents a cornerstone in prostate cancer management [4]. However, patients who undergo endocrine therapy will inevitably progress to castration-resistant prostate cancer. Treatment options at this stage, such as radiotherapy and chemotherapy, are often associated with adverse effects, complications, and poor prognosis. Consequently, developing effective drugs for the comprehensive management of prostate cancer remains a critical focus in medical research.

Traditional Chinese Medicine (TCM) is a unique medical system refined through thousands of years of clinical practice in China. Chinese herbal medicine, an integral component of TCM, serves as a primary therapeutic modality [5]. Recently, many Chinese herbal medicines or extracts have been used to treat or co-treat diseases and have shown good therapeutic effects in clinical studies [6]. Huangqi (Astragalus membranaceus) is a widely used Chinese herb recognized for its diverse pharmacological properties, including immunomodulation, anticancer, lipid-lowering, antihyperglycemic, and antioxidant activities [7]. Zhang et al. found that Huangqi can reduce the viability of prostate cancer cell lines (22Rv1 and LNCaP) and decrease the protein expression level of the androgen receptor (AR) in prostate cancer cells, thereby inhibiting prostate cancer [8]. Through network pharmacology, Lin et al. identified potential targets of Huangqi in prostate cancer treatment, predicting its ability to suppress tumor progression by inducing apoptosis, perturbing the cell cycle, and inhibiting cell migration [9]. Thus, Huangqi represents a significant herbal medicine in prostate cancer therapeutics.

The discovery and application of plant-derived exosome-like nanoparticles (PELNs) have opened new avenues for developing traditional Chinese medicine. Recent studies indicate that PELNs exhibit anti-inflammatory, wound-healing, and anti-tumor properties across various disease models, showing considerable therapeutic potential with minimal observed toxicity [10–12]. Furthermore, the lipid bilayer structure of PELNs protects their cargo from harsh external conditions, making them promising natural drug delivery vehicles. The selection of source plants for PELN extraction generally follows two strategies: one emphasizes safety and convenience by using edible plants like fruits and vegetables, while the other prioritizes therapeutic potency by utilizing medicinal plants. Several Chinese herbal medicines have been used to isolate Chinese herbal medicine-derived exosome-like nanoparticles (CELNs), demonstrating encouraging results in cancer treatment. For instance, ginseng-derived PELNs can alleviate T cell exhaustion in the colorectal cancer microenvironment, enhancing the infiltration of functional CD8+ T cells [13]. In addition, CELNs can also directly cause the death of tumor cells. For instance, Brucea javanica-derived PELNs can induce apoptosis in breast cancer cells via the PI3K/Akt/mTOR pathway [14]. However, the application of CELNs for prostate cancer treatment remains unexplored.

In this study, we isolated Huangqi derived exosome-like nanoparticles (HELNs) from fresh herbal medicine Huangqi (Astragalus membranaceus), and characterized their properties. By integrating single-cell sequencing with in vitro and in vivo experiments, we demonstrated that HELNs reprogram tumor-associated macrophages (TAMs) within the prostate cancer tumor microenvironment (TME) towards an anti-tumor phenotype. Sequencing of the gut microbiota revealed that HELNs can enhance the abundance of antitumor probiotics in mouse guts. In terms of direct effects, HELNs downregulate GPX4 protein expression in prostate cancer cells, inducing ferroptosis. Furthermore, we leveraged HELNs as a drug delivery platform by loading them with siGPX4 to construct siGPX4@HELNs nanoparticles and validated their efficacy both in vitro and in vivo. Collectively, our work comprehensively investigates the therapeutic potential of HELNs against prostate cancer and evaluates their utility as siRNA carriers, providing novel insights into the application of PELNs in cancer therapy.

Materials and methods

Cell culture

22Rv1 cells were provided by Boster (Wuhan, China), RM-1, LNCaP, C4-2 and PC-3 cells were provided by Procell (Wuhan, China), and all five cell lines were authenticated by short tandem repeat (STR) profiling. All five cell lines were cultured in RPMI 1640 medium containing 10% FBS, penicillin (100 U/mL)/streptomycin (100 µg/mL) and maintained at 37 °C in a humidified atmosphere with 5% CO₂.

Isolation and purification of HELNs

Fresh Huangqi roots were mixed with sterile PBS at a 1:4 (g/mL) ratio and homogenized at low speed for 10 min. The homogenate was filtered through a mesh to remove large tissue debris. The filtrate was sequentially centrifuged at 5,000 × g for 30 min and 14,000 × g for 60 min to eliminate cellular debris, organelle fragments, and large particles. The supernatant was then ultracentrifuged at 140,000 × g for 60 min to pellet HELNs. The pellet was resuspended in PBS, filtered through a 0.22 μm filter (without protease or RNase inhibitors). The concentration of the isolated HELNs was measured using the BCA assay, and Aliquots were stored at −80 °C.

Characterization of HELNs

The morphology and structure of HELNs were observed through a transmission electron microscope (TEM) (JEM-1200, JEOL, Japan). The size and concentration of HELNs were characterized by nanoparticle size analysis (NTA, NanoSight NS300, Malvern Panalytical, UK). To measure the Zeta potential, HELNs were diluted to an appropriate concentration, and the measurement was conducted by ZetaView Particle Analyzer (ZetaView PMX120-Z, Particle Matrix, German). NTA and zeta-potential measurements were each performed on three independent batches of HELNs, and the corresponding coefficients of variation were calculated.

Protein and lipid components of HELNs

HELNs were denatured in protein loading buffer at 100 °C for 10 min. A 10% sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) gel was prepared using a PAGE gel rapid preparation kit (Epizyme Biotech, China), and proteins in HELNs were separated by electrophoresis. Subsequently, the protein components were transferred to a PVDF membrane, which was stained with a Coomassie Brilliant Blue kit (Beyotime, China) and then photographed. To detect the lipid components of HELNs, the lipid components were first separated by mixing 200µL HELNs with 800µL methyl tert-butyl ether and 240µL methanol, followed by centrifugation at 14,000 g for 20 min. The upper organic phase containing the lipids was then subjected to separation using the UHPLC Nexera LC-30 A ultra-high-performance liquid chromatography system. After UHPLC separation, the lipids were analyzed by mass spectrometry using the AB 6500 + QTRAP mass spectrometer (AB SCIEX).

Western blot

Protein extraction: RIPA lysis buffer containing protease and phosphatase inhibitors was added to the protein samples, which were then incubated at 4 °C for 30 min to ensure complete lysis. The lysates were subjected to ultrasonication under the following conditions: total duration of 3 min, with each sonication lasting 3 S, intervals of 3 s, and a power setting of 8%. Subsequently, the samples were centrifuged at 13,000 rpm for 15 min, after which the supernatant was collected, mixed with 4⋅ protein loading buffer, and boiled at 100 °C for 10 min. Western blot: a 10% sodium dodecyl sulfate-polyacrylamide gel electrophoresis gel was prepared, and the proteins in the HELNs were separated by electrophoresis. Subsequently, the protein components were transferred to a PVDF membrane. Membranes were blocked with 5% BSA in TBST for 30 min, and were incubated in primary antibody against iNOS, ARG1, β-Tubulin, GPX4 for at 4 °C for 8 h. Then, the membranes were washed by TBST buffer for 3 times and incubated in anti-rabbit IgG or anti-mouse IgG for 1 h at room temperature. After washing 3 times with TBST buffer, the membranes were visualized by an ECL detection system (ChemiDoc ^™ MP, Bio-Rad, USA). The detailed information of antibodies was shown in supplementary material 1.

RT-qPCR

Total RNA of cell samples was extracted using an RNA Extraction Kit (ABclonal, China). RNA reverse transcription was performed using a reverse transcription kit (Yeasen, China) according to the manufacturer’s instructions. Genomic DNA was first removed, and then reverse transcription was carried out following the predefined protocol. Quantitative PCR (qPCR) was performed using SYBR Green Mix (Yeasen, China) in a QuantStudioTM 6 Flex Real-Time PCR machine (Applied Biosystems, USA) under the following steps: denaturation at 95 °C for 5 min, followed by 40 cycles of 95 °C for 15 s, 55 °C for 30 s, and 72 °C for 20 s. After amplification, the instrument automatically analyzes the amplification and melting curves. For each sample group, three technical replicates are used, and the average of the three replicates is taken as the final CT value. When calculating result, the relative gene expression level of the experimental group compared to the control group for a specific target gene is given by 2^-ΔΔCT. The detailed information of primers for different genes used in RT-qPCR is shown in supplementary material 1.

Mouse bone marrow-derived macrophages (BMDMs) Preparation

C57BL/6 mice were euthanized and sterilized in 75% ethanol for 5 min. Subsequently, in a biosafety cabinet, the tibia and femur of the mouse were extracted using sterile forceps. The mouse bone marrow cavity was flushed with sterile PBS, and the obtained bone marrow was centrifuged at 400 g for 5 min. The pellet was treated with red blood cell lysis buffer for 5 min, neutralized, and then centrifuged again at 400 g for 5 min. The pellet was resuspended in complete DMEM medium containing 20ng/mL macrophage colony-stimulating factor (M-CSF, ABclonal, China) and seeded in a 6-well plate, with the day of plating designated as Day 0. Media were changed on Day 3 and Day 5. For M2 polarization, 10 ng/mL IL-4 (ABclonal, China) was added on Day 5. BMDMs were ready for experiments on Day 7.

Animal models and treatment protocol

Six-week-old male C57BL/6 mice and BALB/c nude mice were purchased from Beijing Vital River Laboratory Animal Technology. Animal models used in this study were all subcutaneous transplanted tumor models. To prepare the subcutaneous transplanted tumor model in BALB/c nude mice, 22Rv1 cells in good condition were digested with trypsin, resuspended in PBS, and then mixed with an equal volume of Matrigenel Matrix (ABW, China). BALB/c nude mice were briefly anesthetized with isoflurane, and 22Rv1 cells were inoculated into the flank of the mice. Each mouse was inoculated with 100 µL of the cell suspension containing 5 × 10⁶ 22Rv1 cells. Before inoculating C57BL/6 mice, depilatory cream was used to remove hair from the flank. The mice were briefly anesthetized with isoflurane, and 100 µL of a suspension containing 1 × 10⁶ RM-1 cells was injected into the flank of each mouse.

After tumor inoculation, mice were allocated to groups by random numbers generated with software. Tumor measurements were performed under blinding: the investigator who measured the longest and shortest diameters was unaware of the group assignments, eliminating subjective bias. Tumor volume was measured using caliper and estimated as length × width × width/2. When tumor volume exceeded 150 mm³, treatment was initiated. Before treatment, the mice were weighed, and the length and width of the tumor were recorded. The first day of treatment was designated as Day 0, with a dosage of 3 mg/kg based on the mouse’s body weight. Each mouse received a 100 µL injection of HELNs at their respective doses, with the injection procedure standardized to a duration of 5 s to ensure consistent delivery. The treatment was administered every 3 days for a total of three doses. On Day 9, the mice were weighed and the length and width of the tumor were recorded once more before being euthanized via CO₂ anesthesia.

Fecal microbiota transplantation (FMT) and hemolysis assay

Donor fecal slurry was prepared by aseptically collecting fresh feces from three groups of mice: healthy, tumor-bearing, and tumor-bearing mice treated with HELNs via intraperitoneal injection. The fecal pellets were homogenized in sterile PBS, centrifuged at 1,000 × g for 1 min, and the supernatant was collected as the FMT inoculum. Recipient mice were first depleted of their endogenous gut microbiota by daily oral gavage of an antibiotic cocktail (ampicillin, neomycin sulfate, metronidazole, and vancomycin) for five consecutive days. After a two-day washout period, FMT was initiated, during which recipient mice received 200 µL of the donor fecal supernatant via oral gavage daily for two weeks. Following this FMT period, recipient mice were subcutaneously inoculated with RM-1 cells, and tumor growth was monitored with the initial measurement taken seven days post-inoculation.

Hemolysis assay: A hemolysis assay was performed to evaluate the biocompatibility of HELNs. Whole blood was collected from healthy mice and centrifuged at 3,000 rpm for 15 min. The pellet containing red blood cells (RBCs) was washed repeatedly with PBS until the supernatant became colorless. The washed RBCs were then resuspended in PBS. Subsequently, the RBC suspension was mixed with HELNs solution at various concentrations in a 1:1 volume ratio. The negative control consisted of RBCs resuspended in PBS alone, while the positive control consisted of RBCs treated with 0.2% Triton X-100. All mixtures were incubated at 37 °C for 1 h, followed by centrifugation at 3,000 rpm for 15 min. Hemolysis was assessed by observing the presence of hemoglobin in the supernatant and quantified by measuring the absorbance at 450 nm.

Single cell RNA sequencing

A subcutaneous transplanted tumor model was established in C57 BL/6 mice. Mice were randomly divided into two groups: a control group and a HELN treatment group (n = 3). Treatment was administered according to the animal experiment protocol, and tumors were excised at the end of the experiment. The dissociation of tumor tissue, cell preparation, single cell RNA-seq library construction and sequencing are shown in supplementary material 1.

The scRNA-seq data utilized in this study were analysis using R with package Seurat. Seurat objects were generated based on mouse scRNA-seq data. The raw gene-expression matrices were processed using the package DoubletFinder and the doublet were filter. Then, cell count matrix were processed to filter for cells with mitochondrial RNA content less than 20%, the number of detected genes fewer than 300 or higher than 7,000, UMI counts fewer than 10,000. Following the preprocess, 47,893 cells were detected with high quality. To estimate the effects of cell cycle heterogeneity, the expression of S or G2/M phases genes was scored in each cell with CellCycleScoring function. And the differences of cell cycle score were regressed out. To integrate the six samples, 3,000 integration features were used and applied Harmony for batch correction. Unsupervised clustering was performed using FindClusters function (resolution = 0.8). Both marker genes analyzed by package COSG (n_gene = 200) and marker gene published in reliable research were used in cell identification. The tumor cell identity was confirmed by inference of copy number variations with packages infercnv and copykat. All the immune cell clusters were specified as a normal reference.

The clusterProfiler package was used in the gene functions and pathway enrichment analysis. All the p-values were adjusted by Bonferroni method and a threshold of 0.05 was set for filtering enrichment results. The top 200 differential expressed genes of each cluster were analyzed through the Gene Ontology (GO) terms, Kyoto Encyclopedia of Genes and Genomes (KEGG) pathways and Reactome Knowledgebase. To quantify the potential cell communication networks, the package CellChat was used. The result was calculated using CellChatDB of mouse and adjusted with the netSmooth package using mouse protein-protein interactions. The visualization of cell communications was finished using the CellChat package.

Gut microbiota sequencing

When animal experiment was concluded, the mice were euthanized. The abdominal cavity of the mice was swiftly opened with surgical scissors, the colon was cut open, and the feces were removed, ensuring that at least three intact fecal pellets were collected from each mouse. After removal, the feces were quickly frozen in liquid nitrogen and then stored at −80 °C. DNA extraction, amplification, library construction and sequencing are shown in supplementary material 1.

The alpha and beta diversity analysis was finished in R with package phyloseq and vegan. The microbial diversity in samples was estimated using the alpha diversity that include Shannon index, Chao1 index, goods_coverage, Simpson index. The bray_curtish distance and Jaccard distance were calculated for Principal coordinates analysis (PCoA) to estimate the beta diversity. Then, the R package microeco was used to analyze the significant differences between two groups. The linear discriminant analysis effect size (LEfSe) method was also used to compare the taxonomy abundance spectrum.

Bulk RNA sequencing

22Rv1 cells were randomly divided into two groups: a control group and a HELN treatment group (n = 3). 24 h after HELNs intervention, total RNA of cells was extracted using TRIzol (Invitrogen, USA). Subsequently, the extracted RNA was subjected to quality assessment, library construction, and sequencing. The obtained results were analyzed using R. DESeq2 was employed to analyze differentially expressed genes (DEGs) between the experimental and control groups, with genes satisfying a q-value < 0.05 and a fold change > 2 or < 0.5 being defined as DEGs. Based on the experimental setup, DEGs were filtered using these thresholds, and the filtered DEGs were subjected to GO, KEGG, Reactome, and WikiPathways enrichment analyses to identify significantly enriched pathways and functions. GSEA analysis was also performed.

In vitro cellular uptake of HELNs

Cells were seeded in 6-well plates one day in advance (4 × 10⁵/well). PKH26 dye (Maokang bio, China) was used to label the HELNs. Briefly, 5 µL PKH26 was mixed with 200 µL Diluent C, combined with 100 µL HELNs in 200 µL Diluent C, and incubated for 10 min at room temperature. An equal volume of fetal bovine serum (FBS) was added for neutralization, followed by centrifugation at 140,000 g for 30 min to obtain the labeled HELNs. After washing with PBS and centrifuging again, the stained HELNs were added to the cells. Flow cytometry was performed at 0 h, 6 h, 12 h, and 24 h to detect the uptake efficiency of the cells. The 0 h group involved adding HELNs to the cells for 5 min before detection to observe the immediate uptake of HELNs by the cells.

Cell viability, colony formation and transwell assays

Cell viability was assessed using CCK-8 assay kit (ABclonal, China). Cells were seeded in 96-well plates (9000 cells/well) and cultured in an incubator for 12 h to allow cell attachment. Different concentrations of the intervention agents were added according to experimental requirements. After 48 and 72 h of intervention, the assay was initiated. The intervention agents were first removed, and each well was washed with 100 µL PBS to prevent residual HELNs and cell debris from affecting the results. A serum-free culture medium containing 10% CCK-8 reagent was prepared, and 100 µL of this medium was added to each well, followed by incubation at 37 °C for 2 ~ 3 h. Cell viability was calculated by measuring the absorbance at a wavelength of 450 nm.

Prostate cancer cells were seeded in 6-well plate, different concentrations of the intervention agents were added according to experimental requirements. After 24 h of intervention, cells were trypsinised and seeded in 6-well plate (800 cells/well). The complete medium was changed regularly. After 2 ~ 3 weeks, colonies were stained with 1% crystal violet and counted. Transwell assay was conducted under the same intervention conditions. After intervention, cells were digested, resuspended in serum-free medium, and added to the transwell inserts (80,000 cells/well), which were then placed in 24-well plates (with 600 µL of complete medium containing 20% FBS). After 48 h of incubation in the incubator, the inserts were removed, and the cells were stained with crystal violet for 30 min. Each experimental group included three biological replicates in every assay. The cells that did not migrate were wiped off, and the migrated cells were observed, photographed, and counted under a microscope. Due to their low migratory and invasive capacities, the transwell assay results for C4-2 and LNCaP cells were not included.

ROS, Fe²⁺ and lipid peroxide detection

Intracellular reactive oxygen (ROS), Fe²⁺ and lipid peroxide were labeled by ROS detective kit (Beyotime Biotechnology, China), FeRhoNox-1 (MCE, USA) and C11 BODITY (Beyotime Biotechnology, China) respectively, and were detected by flow cytometry. Specifically, prostate cancer cells were seeded in 6-well plate (4 × 10⁵/well) and treated with HELNs. After 24 h, HELNs were removed; different detection probes were diluted with serum-free medium and added into 6-well plate. After incubation at 37 °C for 30 min, the probes were removed, and the cells were washed three times with PBS before being digested. Fluorescence intensity was then detected using a flow cytometer, and each group contained three biological replicates.

Flow cytometry

Apoptosis: prostate cancer cells were seeded in 6-well plates (4 × 10⁵/well) and subjected to interventions according to the experimental schedule. Prostate cancer cells were digested after intervention and stained with an apoptosis kit (KeyGEN, China) according to the manufacturer’s instructions, followed by detection of the apoptotic ratio using flow cytometry.

Macrophages: The tumor sample was minced and placed in serum-free medium containing DNase, hyaluronidase, and collagenase at concentrations of 0.1 mg/mL, 0.1 mg/mL, and 1 mg/mL, respectively. The minced tissue was then incubated at 37 °C for 90 min to digest the sample until cells were released. After digestion, the tumor tissue was gently homogenized on ice, and the cell suspension was filtered through a 74 μm pore-sized mesh to remove undigested tissue. The cell suspension was subjected to density gradient centrifugation to isolate immune cells, with tumor cells being discarded. The isolated cells were lysed to remove red blood cells and stained with Zombie dye to distinguish live and dead cells. Subsequently, cells were stained for surface proteins, fixed and permeabilized, and stained for intracellular proteins. After staining, cells were resuspended in PBS, filtered through a 74 μm pore-sized mesh, and analyzed by flow cytometry (CytoFLEX, Beckman, USA). Information on the antibodies used for flow cytometry is presented in supplementary material 1.

Immunohistochemical staining and H&E staining

Prior to staining, tumor tissues or mouse organs were embedded in paraffin.

H&E staining: The tissue sections were placed in hematoxylin staining solution for 5 min, then removed and rinsed with ddH2O. Subsequently, they were treated with differentiating solution, rinsed again with ddH₂O, and then subjected to bluing treatment before being thoroughly washed with running water. After dehydration, the sections were stained with eosin dye. Finally, after dehydration and mounting, the sections were photographed.

Immunohistochemical staining: The embedded tumor tissue was dewaxed, followed by antigen retrieval with EDTA antigen retrieval solution (RecordBio, China) for 35 min. After washing the sections with PBS, peroxidase was blocked. Subsequently, the sections were blocked with 3% BSA and incubated sequentially with primary and secondary antibodies. DAB chromogenic solution (RecordBio, China) was used for staining, and hematoxylin was used for nuclear staining. The stained sections were then dehydrated, mounted, and photographed. The antibodies used for immunohistochemistry are the same as those used for Western blot. The immunohistochemical results were analyzed using ImageJ. In brief, after deconvolution of the immunohistochemistry outcomes, the grayscale values were converted into optical density (OD) values. The regions of expression were delineated through a combination of automated system detection and manual outlining, and the OD values were then measured to reflect the protein expression levels.

Transfection of siGPX4

siRNA sequences used in this study are presented in supplementary material 1. Prostate cancer cells were seeded in 6-well plates at a density of 3 × 10⁵ cells per well using antibiotic-free medium one day prior to transfection. For transfection, siRNA at a concentration of 50 nM was added to each well in serum-free medium containing 6 µL of INTERFERin^® reagent (Polyplus, France). The plates were then incubated at 37 °C for 6 h. After transfection, the medium was replaced with complete medium, and the expression levels of the corresponding genes were detected at the RNA and protein expression levels 48 h later.

Reversible electroporation

Reversible electroporation was performed in an electroporation cuvette with the electroporation volume set at 800 µL. For each electroporation, 768 µg of HELNs and 160 nmol of siGPX4 were added to the cuvette, followed by sterile PBS to bring the volume up to 800 µL. The cuvette was then sealed and placed in the electroporator (ECM830, BTX, USA) with the following conditions: voltage set at 400 V/cm, discharge interval at 0.1 s, and number of discharges at 3 times. After electroporation, the cuvette was placed in a refrigerator at 4 °C for 3 h to allow the fusion of gaps in the HELNs caused by electroporation.

In vivo biodistribution of HELNs

According to the manufacturer’s instructions, HELNs were labeled using a DiR-Membrane Labeling and Purification Kit (DLM, China). A mixture of 50 µL HELNs, 50 µL reaction solution, and 5 µL DiR dye was thoroughly combined and incubated in the dark at 37 °C for 30 min. The incubated DiR-labeled HELNs mixture was then applied to a centrifuge column and centrifuged at 100 × g for 90 S. The filtrate in the collection tube was removed, and after adding PBS, the mixture was centrifuged again. The filtrate in the collection tube contained the DiR-labeled HELNs.

Subcutaneous transplanted tumors were established in C57BL/6 mice as described above. When the tumor size reached 200 mm³, DiR-labeled HELNs were administered to the mice via intraperitoneal injection. At 24 h, 48 h, and 72 h post-injection, the mice were euthanized under CO₂ anesthesia, and the heart, liver, spleen, lung, kidney, tumor, and intestine were removed for imaging. Imaging was performed using the IVIS Spectrum Imaging System (PerkinElmer, USA). The excitation wavelength was set at 745 nm, and the emission wavelength was 800 nm. Image acquisition and analysis were facilitated by Living Image^® software (PerkinElmer, USA).

Statistical analysis

Quantitative data are presented as mean±SD. For normally distributed values between two groups, a two-tailed Student’ t-test was used. For non-normally distributed variables, the Mann-Whitney U test was applied. For comparisons among multiple groups, analysis of variance (ANOVA, Tukey‘s Multiple Comparisons test) was used. All statistical analyses were performed using GraphPad Prism (version 9.0). A p-value less than 0.05 was considered statistically significant.

Results

Isolation and characterization of Huangqi derived exosome like nanoparticles (HELNs)

HELNs were isolated from fresh Huangqi (Astragalus membranaceus) root using a differential centrifugation protocol (Fig. 1A). Transmission electron microscopy (TEM) confirmed their spherical morphology (Fig. 1B). Nanoparticle tracking analysis (NTA) showed that HELNs have a relatively narrow size distribution with a peak at 112 nm (Fig. 1B). The stability of HELNs was also assessed via NTA and zeta potential measurements under the following conditions: after storage at −80 °C for two months (Fig. 1C), and after 24 h incubation in acidic (pH 4, Fig. 1D) and alkaline (pH 10, Fig. 1E) buffers. The corresponding zeta potential data are presented in Fig. 1F. The batch-to-batch consistency of HELNs was assessed using the coefficient of variation (CV). The CV for nanoparticle size (from NTA), zeta potential and protein concentration were calculated to be approximately 1.34%, 1.87% and 6.66%, respectively. Both SDS-PAGE and proteomic analyses indicated that the protein components of HELNs are predominantly distributed within a molecular weight range of 10 to 43 kDa (Fig. 1G, H). Lipidomic profiling identified the major lipid constituents as Ceramide (Cer, 29.7%), Phosphatidic Acid (PA, 16.2%), Hexosylceramide (Hex1 Cer, 11.58%), Triacylglycerol (TG, 9.4%), and Phosphatidylethanolamine (PE, 7.5%) (Fig. 1I). This lipid profile is distinct from those of other PELNs previously reported. Cer is an amide compound formed by the dehydration of long-chain fatty acids and sphingosine, playing a crucial role in the composition of cell membranes, especially lipid rafts, and mediating protein-lipid interactions [15]. Trajkovic et al. demonstrated that Cer is involved in the formation of multivesicular bodies during the exosome secretion process, occupying a key regulatory position and being enriched in purified exosomes [16]. The high content of Cer in HELNs suggests that they are likely secreted via the multivesicular body pathway.

Fig. 1.

Extraction process and characterization of HELNs. (A) Schematic illustration of HELNs isolation. (B) Particle size distribution showing a main peak at 112 nm (NTA), TEM image revealing spherical morphology. (C) Particle size distribution of HELNs after storage at −80 °C for two months. (D) Particle size distribution of HELNs after incubation in acidic solution (pH 4.0) for 24 h. (E) Particle size distribution of HELNs after incubation in alkaline solution (pH 10.0) for 24 h. (F) Zeta potential of HELNs. Measurements were performed under four conditions: freshly prepared, storage at −80 °C for two months, incubation at pH 4.0 for 24 h, and incubation at pH 10.0 for 24 h. (G) Protein distribution in HELNs (Coomassie staining). (H) Protein distribution (kDa) in HELNs. (I) Composition and proportion of top 10 classes of lipids from HELNs

In vitro uptake and tumor-suppressive effects of HELNs

Flow cytometry analysis demonstrated efficient time-dependent uptake of PKH26-labeled HELNs by four prostate cancer cell lines (22Rv1, LNCaP, PC-3, C4-2) at 0 h, 6 h, 12 h, and 24 h (Fig. 2A). C4-2 exhibited the fastest uptake, reaching 98% at 12 h, followed by PC-3 and LNCaP, while 22Rv1 had the slowest uptake rate. To confirm that HELNs were internalized, as opposed to merely adsorbed onto the cell surface, we performed confocal imaging of 22Rv1 and PC-3 cells following incubation with HELNs (Figure S1A, B). Subsequently, we examined the cytotoxic effects of HELNs on prostate cancer cells using the CCK-8 assay to measure cell viability following treatment with varying concentrations of HELNs. HELNs demonstrated robust time- and dose-dependent cytotoxicity against the four prostate cancer cell lines (Fig. 2B). Notably, LNCaP and C4-2 cells were particularly sensitive, showing significant viability reduction at 8 µg/mL. Overall, cell viability was reduced to approximately 50% at concentrations of 32 to 64 µg/mL. Therefore, for subsequent experiments, 32 µg/mL was selected as the low-concentration group and 64 µg/mL as the high-concentration group for treating prostate cancer cells. We also evaluated HELNs cytotoxicity against the non-malignant prostatic epithelial cell line RWPE1 (Figure S1C). Viability remained markedly higher than that of prostate-cancer cells at 48 h, and a moderate decline was observed at 72 h.

Fig. 2.

Cellular uptake of HELNs by prostate cancer cells and the cytotoxic effects of HELNs on prostate cancer cells in vitro. (A) Uptake efficiency of four prostate cancer cells (22Rv1, LNCaP, C4-2 and PC-3) of PKH26 labeled HELNs for 0, 6, 12 and 24 h, analyzed by flow cytometry. (B) Cytotoxic effects of HELNs on prostate cancer cells detected by CCK-8 assay. (C) HELNs induce apoptosis in prostate cancer cells in a concentration-dependent manner. (D) HELNs inhibit colony-forming ability of 22Rv1, C4-2 and PC-3 cells. (E) HELNs inhibit transwell migration capacity of 22Rv1 and PC-3 cells. *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001, ns no significance

Many PELNs have been reported to induce apoptosis in tumor cells [17–19]. Cells were divided into three groups: a control group and two intervention groups with concentrations of 32 µg/mL and 64 µg/mL. HELNs induced concentration-dependent apoptosis in all four cell lines (Fig. 2C). Furthermore, HELNs significantly inhibited colony formation and Transwell migration of prostate cancer cells (Fig. 2D, E).

We next evaluated the optimal administration route for HELNs in vivo (Figure S1D). The administration route for PELNs depends on the target disease, with intraperitoneal injection, gavage, and intravenous injection being common methods, all of which have been reported in tumor treatment [20]. Subcutaneous tumor-bearing mice were treated with HELNs via intraperitoneal (i.p.), intravenous (i.v.), or intragastric (i.g.) administration. The results showed that i.g., i.p. and i.v. of HELNs significantly reduced tumor volume and tumor weight (Figure S1E-H). However, i.v. injection caused a decrease in body weight by Day 9 (Figure S1I, J). This suggests that intravenous injection may have a potential impact on mouse health. To comprehensively evaluate the safety profile of HELNs, we performed a hemolysis assay and conducted H&E staining of major organs from mice following intravenous administration. The results demonstrated that HELNs induced negligible hemolysis even at a high concentration of 128 µg/mL (Figure S1K, L). Furthermore, the organ sections revealed no significant pathological alterations, indicating that intravenous injection of HELNs did not adversely affect the major organs (Figure S1M). The observed reduction in mouse body weight warrants further investigation into its underlying cause. Hence, intraperitoneal injection was adopted as the primary administration method for HELNs in the following experiments.

Single-cell sequencing revealed the immune landscape after HELNs intervention

Single-cell sequencing is an unbiased, high-throughput, and high-resolution transcriptomic analysis performed on individual cells, capable of effectively reflecting the infiltration status of tumor and immune cells in TME [21, 22]. To elucidate the effects of HELNs on the prostate cancer TME post-treatment at single-cell resolution, we collected 6 subcutaneous transplanted tumor samples (3 from HELNs group, 3 from control group) from 6 mice. After preprocessing and quality control (Figure S2A), we obtained 36,424 high quality cells of 20 clusters (Figure S2B, C). Subsequently, cell annotation was performed based on specific genes and generated detailed cell identity annotation: neutrophils, M1 macrophages, M2 macrophages, epithelia, fibroblasts, B cells, T cells (Fig. 3A, B). HELNs treatment increased the overall proportion of immune cells within the TME, with notable expansions in neutrophils and M1 macrophages (Fig. 3C and Figure S2D).

Fig. 3.

Single-cell sequencing reveals that HELNs remodel the intratumoral immune landscape. (A) Overview of major cell types of 36,424 cells from HELNs intervention group (n = 3) and control group (n = 3). (B) Dotplot for the marker genes in each cell type. (C) Normalized cell type composition stacked bar chart by groups. (D) Heatmap for the number and strength of interactions among all cell types in samples. (E) Bar chart for the number of macrophages by groups. (F) Normalized cell type composition stacked bar chart in macrophage cells by groups. (G) Dotplot for the functional genes in M1 macrophages by groups. The size of the circle represents the proportion of cells expressing the relevant gene within that cell type, with larger circles indicating a higher proportion. Color blue represents the level of average expression, with darker shades of blue indicating higher expression levels. (H) KEGG gene enrichment analysis between HELNs intervention group and control group. (I) Dotplot for lipids metabolism strength in macrophages by groups.

To identify the key immune cells in prostate cancer TME that respond to and exert effects following HELNs intervention, we conducted an analysis of cell-cell communication strength. Fig. 3D illustrates the changes in the number (left) and strength (right ) of communications between different cells following HELNs intervention compared to the control group. The vertical axis represents the initiating cells of communication, while the horizontal axis represents the receiving cells. Red indicates an increase in communication strength, and blue indicates a decrease. Although the proportion of neutrophils increased, their interaction strength with other cells decreased post-HELNs treatment. In contrast, communication involving M1 macrophages was enhanced after HELNs intervention, suggesting an increase in their functionality. The detailed signal pathways and the patterns upgrading after HELNs treatment were displayed in Figure S2E. HELNs significantly increased the total number of macrophages (Fig. 3E), with the majority being M1 macrophages, as evidenced by a significant increase in their proportion (Fig. 3F).

M1 macrophages play a role in tumor development by presenting antigens and phagocytosing tumor cells, thereby suppressing tumors. The expression levels of key genes in macrophages involved in anti-tumor immunity was analyzed (Fig. 3G). The results showed that the proteins iNOS, IL-12 and TNF-α (corresponding genes: Nos2, Il12 and Tnf), which are associated with the antitumor functions of M1 macrophages significantly upregulated following HELNs intervention. DEGs analysis was conducted in macrophages between HELNs intervention group and control group, with absolute value of logFC > 0.5 and p < 0.05 (Figure S2F). The top 200 differential expressed genes of each cluster were analyzed through the GO, KEGG pathways and Reactome Knowledgebase. The enrichment pathways showed significant changes in inflammatory, immunity and metabolism (Figure S2G, H). As Fig. 3H showed, KEGG revealed that M1 macrophages were enriched in the IL-17 signaling pathway, TNF signaling pathway and lipid metabolism pathways. Following the guidance of the KEGG enrichment analysis result, we identified the metabolic landscape of macrophages. The result showed that HELNs influenced the lip metabolism of M1 macrophages especially in increasing fatty acid degradation (Fig. 3I).

In summary, scRNA-seq revealed that HELNs remodel the prostate cancer TME, particularly by enhancing the proportion and functional state of anti-tumor M1 macrophages.

HELNs reverse the M2-like polarization of macrophages and induce M1-like polarization

Guided by scRNA-seq findings, we investigated HELNs’ effects on macrophage polarization in vitro and in vivo. We isolated BMDMs from mice and cultured them under different conditions to induce them into undifferentiated M0 macrophages and M2 macrophages. We observed that when a low concentration of HELNs was added, macrophages exhibited significant morphological changes, indicating that a low concentration (5 µg/mL) of HELNs can alter the polarization state of macrophages. Therefore, we used 5 µg/mL (Low C.) as the low concentration and 10 µg/mL (High C.) as the high concentration to treat macrophages. In M0 macrophages, HELNs significantly increased M1 marker gene expression (Nos2, Cd80, Tnfa; Fig. 4A) and slightly increased Arg1 while decreasing Cd206 (M2 marker; Fig. 4B). In M2-polarized macrophages, HELNs significantly increased the expression of M1 polarization-related genes and significantly reduced the expression of M2 polarization-related genes (Fig. 4C, D). Western blotting confirmed HELNs increased the expression of iNOS (gene name Nos2) and ARG1 in M0 macrophages. At this point, the slight upregulation of ARG1 expression is a normal manifestation of macrophage functional activation, as this expression can also be detected in IFNγ-induced M1 macrophages. HELNs significantly reduced the expression of ARG1 in M2 macrophages while increasing the expression of iNOS (Fig. 4E).

Fig. 4.

HELNs reverse M2-like polarization and induce M1-like polarization of macrophage. (A) mRNA expression levels of M1 polarization markers (Nos2, Cd80 and Tnfa) and (B) M2 polarization markers (Arg1, Cd206 and Fizz1) in M0 macrophages after HELNs treatment. (C) mRNA expression levels of M1 and (D) M2 polarization markers in M2 macrophages after HELNs treatment. (E) Protein expression levels of iNOS and ARG1 in M0 and M2 macrophages after HELNs treatment. (F) Schematic diagram of animal experiment. (G) Photo of tumors at the end of experiments (n = 5). (H) Tumor growth profiles in tumor-bearing mice. (I) Tumor volume at the end of experiments. (J) Tumor weight at the end of experiments. (K) Body weight changes of tumor bearing mice. (L) Percentage of ARG1, CD206, iNOS, and CD80-expressng macrophages from tumor area in control and HELNs-treated groups, detected by flow cytometry. (M) Comparison of CD206⁺ and ARG1⁺ cell proportions between two groups. (N) Comparison of iNOS⁺ and CD80⁺ cell proportions between two groups. (O) Immunofluorescence characterization of tumor-infiltrating macrophages in control and HELNs-treated groups, DAPI (blue), F4/80 (yellow), CD80 (red), CD206 (green). *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001

We also validated the effects of HELNs on macrophage polarization in the TME in vivo. A subcutaneous transplanted model of prostate cancer was established in C57BL/6 mice. A total of three doses were administered over a period of 9 days (Fig. 4F). HELNs significantly reduced the tumor burden in mice without affecting their body weight (Fig. 4G-K), which is consistent with the findings in the earlier sections of this study. Tumor tissues were dissociated, and immune cells were isolated for flow cytometric analysis. Macrophages were defined as the cell population positive for CD45, CD11b, and F4/80 (Figure S3A). The results revealed that the proportion of M1 macrophages among total macrophages was significantly higher in HELNs-treated group compared to control group, while the proportion of M2 macrophages was significantly reduced after treatment (Fig. 4L-N). Immunofluorescence staining of tumor tissue sections also showed consistent results (Fig. 4O).

The safety of HELNs in vivo was also evaluated. At the end of the animal experiments, major organs (heart, liver, spleen, lung, kidney) of mice in both groups were harvested and stained with hematoxylin and eosin (H&E). The results showed that HELNs had no impact on the morphology and structure of the organs (Figure S3B). Considering the potential hepatotoxicity and nephrotoxicity associated with traditional Chinese medicine, blood samples were collected from the mice after treatment to measure liver and kidney function indicators. The results indicated that HELNs did not significantly alter these indicators (Figure S3C), preliminarily suggesting that HELNs are safe for in vivo application.

HELNs increase the abundance of anti-tumor probiotics in mouse gut

To clarify the in vivo biodistribution of HELNs, HELNs were stained with DiR dye. Organs of C57BL/6 mice with subcutaneous transplanted tumors were imaged at 24 h, 48 h, and 72 h after intraperitoneal injection of DiR-stained HELNs. The imaging results indicated that HELNs were primarily concentrated in the gut, liver, and spleen, with some also accumulating in the tumor area (Figure S4A). This result suggests that HELNs may potentially influence the gut microbiota. Previous studies have demonstrated that PELNs can alter the abundance of gut microbiome and exerts anti-tumor effects [23, 24]. To further investigate the influence of HELNs, we analyzed the alterations in gut microbiota after treatment with HELNs.

A total of eight samples were incorporated into the final analysis. As Fig. 5A showed, Bacteroidota, Firmicutes, Desulfobacterota and Proteobacteria were the most common abundant at the phylum level. The differences of most abundant 25 genera were showed in Figure S4B. Alpha diversity indices (Ace, Shannon, Chao1, goods_coverage, Simpson) showed no significant difference between groups (Fig. 5B), indicating that HELNs did not alter overall microbial richness. However, beta diversity analysis (Bray-Curtis, Jaccard) revealed significant compositional shifts (Fig. 5C, D). To further investigate the microbial differences between the HELNs treatment group and control group, an analysis employing linear discriminant analysis effect size (LEfSe) unveiled distinct compositions (p < 0.05) in Fig. 5E. The results indicated that the abundance of Verrucomicrobia, Verrucomicrobiae, Marinifilaceae Verrucomicrobiales, Akkermansiaceae, Akkermansia and Odoribacter was increased following HELNs intervention. The genus Akkermansia comprises only three species, among which Akkermansia muciniphila is the most extensively studied (Fig. 5F). Akkermansia muciniphila has been reported to exert anticancer effects in various solid tumors, including lung cancer and colorectal cancer [25, 26]. In prostate cancer, Akkermansia muciniphila can sensitize prostate cancer to androgen deprivation therapy while modulating the TME [27, 28]. In general, intraperitoneal injection of HELNs specifically enriches beneficial anti-tumor gut bacteria without disrupting overall microbial diversity.

Fig. 5.

HELNs exert anti-tumor effects by altering the gut microbiota landscape in C57BL/6 mice. (A) Stacked bar chart of microbiome in HELNs-treated group and control group at phylum level. (B) Alpha diversity of microbiome, including Shannon index, Simpson index, observed_species, Chao, Ace, and goods_coverage. (C) Bray-Curtis distance of microbiome by groups. (D) PCoA of profiling data from the gut microbiota in HELNs-treated group and control group (based on Bray and Jaccard distance) (E) LEfSe analysis including bacteria that were significantly different in relative abundance between HELNs-treated group and control group. (F) The LEfSe result showing in Cladogram. (G) Schematic diagram of the fecal microbiota transplantation (FMT) experiment. (H) Tumor growth profiles in tumor-bearing mice. (I) Photo of tumors at the end of experiments (n = 4). (J) Tumor weight at the end of experiments. (K) Tumor volume at the end of experiments. (L) Body weight changes of tumor bearing mice. *P < 0.05, ns no significance

To establish a causal relationship and further demonstrate that the alteration of gut microbiota by HELNs contributes to their antitumor effects, we performed fecal microbiota transplantation (FMT) experiments (Fig. 5G). Mice in different experimental groups received fecal slurries derived from donor mice subjected to one of three conditions: healthy, tumor-bearing, tumor-bearing mice which treated with HELNs. The results showed that FMT from tumor-bearing mice promoted tumor growth, whereas FMT from HELNs-treated mice significantly reduced the tumor burden (Fig. 5H-K). Throughout the experiment, no significant differences in body weight were observed among the groups (Fig. 5L).

HELNs induce ferroptosis of prostate cancer cells

Through integrated multi-omics analysis, we determined that HELNs exert anti-tumor effects by influencing immune cells in the TME and the gut microbiota. Subsequently, we explored the mechanism by which HELNs directly kill prostate cancer cells. Bulk RNA-seq of 22Rv1 cells treated with HELNs revealed significant alterations in gene expression (Figure S5A, B). Pathway enrichment analysis of differentially expressed genes (DEGs) highlighted significant induction of ferroptosis-related pathways (Fig. 6A, B) and oxidative stress responses (Figure S5C). TEM images of 22Rv1 and LNCaP cells revealed typical signs of ferroptosis after HELNs intervention (Fig. 6C, red box: increased mitochondrial electron density, mitochondrial shrinkage, and disruption of mitochondrial cristae). Flow cytometry was subsequently used to further confirm the occurrence of ferroptosis. Lipid peroxidation and iron overload are two hallmark biochemical changes in the ferroptosis process and are the means by which we detect ferroptosis [29, 30]. Cells were divided into three groups: a 5% PBS group, a 32 µg/mL HELNs intervention group, and a 64 µg/mL HELNs intervention group. Flow cytometry confirmed HELNs concentration-dependently increased intracellular ROS (Fig.ure 6D, E), lipid peroxidation (Fig. 6F, G), and Fe²⁺ levels (Fig. 6H, I) in 22Rv1 cells (similar results in LNCaP, Figure S5D-F). Meanwhile, the effect size of HELN-induced ferroptosis was quantified by computing Cohen’s d for the mean fluorescence intensity of lipid peroxidation. The effect size for HELN-induced lipid peroxidation was substantial (Cohen’s d: 22Rv1 = 10.38, LNCaP = 92.36). These findings clearly indicate that HELNs induce ferroptosis in prostate cancer cells. GPX4 is the only known mammalian enzyme that can reduce phospholipid hydroperoxides to their corresponding phospholipids, preventing unnecessary ferroptosis, thus playing a crucial role in the ferroptosis process [30]. Western blot showed HELNs downregulated GPX4 protein expression in a concentration-dependent manner (Fig. 6J), consistent with ferroptosis induction. Furthermore, we examined the expression of other ferroptosis-related proteins in both cell lines (Figure S5G, H). The results indicate that HELNs treatment did not alter the expression levels of SLC7A11 or ACSL4. Notably, however, a slight reduction in FSP1 expression was observed.

Fig. 6.

HELNs induce ferroptosis of prostate cancer through downregulating the expression level of GPX4. (A) KEGG enrichment analysis results of 22Rv1 cells after HELNs treatment. (B) WikiPathways enrichment analysis results of 22Rv1 cells after HELNs treatment. (C) TEM images of 22Rv1 and LNCaP cells in control and HELNs-treated groups (red arrows indicate mitochondria with typical ferroptosis-related changes and normal mitochondria). (D) Flow cytometry detecting the intracellular ROS levels of 22Rv1 cells. (E) Quantification of ROS levels in 22Rv1 and LNCaP cells by mean fluorescence intensity (MFI). (F) Flow cytometry detecting the intracellular peroxidized lipid levels of 22Rv1 cells. (G) Quantification of peroxidized lipid levels in 22Rv1 and LNCaP cells by MFI. (H) Flow cytometry detecting the intracellular Fe²⁺ levels of 22Rv1 cells. (I) Quantification of Fe²⁺ levels in 22Rv1 and LNCaP cells by MFI. (J) Protein levels of GPX4 in 22Rv1 and LNCaP cells in different groups. (K) Schematic diagram of animal experiment. (L) Photo of tumors at the end of experiments (n = 5). (M) Tumor growth profiles in tumor-bearing mice. (N) Tumor volume at the end of experiments. (O) Tumor weight at the end of experiments. (P) Body weight changes of tumor bearing mice. (Q) Protein expression levels of GPX4 in tumor areas, detected by immunohistochemical staining. *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001

To verify the in vivo cytotoxic effects of HELNs on human prostate cancer cells, this study subsequently inoculated 22Rv1 cells in BALB/c nude mice to establish a subcutaneous xenograft tumor model (Fig. 6K). HELNs significantly inhibited tumor growth (Fig. 6L-N), and reduced tumor weight (Fig. 6O). The body weight of the mice did not change significantly during the intervention period (Fig. 6P). Immunohistochemical staining confirmed decreased GPX4 expression in HELNs-treated tumors (Fig. 6Q, Figure S5I). Subsequently, the safety of HELNs in vivo was re-evaluated. The results showed that after HELNs intervention, there were no significant differences in liver and kidney function indicators in the blood biochemistry of mice compared with the control group (Figure S5J). There were no significant morphological and structural differences in the main organs of mice compared with the control group (Figure S5K).

HELNs loaded with siGPX4 induce ferroptosis in prostate cancer cells

As bilayer lipid membrane-encapsulated vesicles, PELNs can not only function as drugs themselves but also serve as carriers to deliver specific drugs across biological barriers and facilitate cellular internalization in vivo [31]. Compared to artificially synthesized drug carriers, PELNs can remain in the body for a longer period and exhibit good cross-species safety and low immunogenicity [32]. We attempted to use HELNs as drug delivery vehicles to further enhance their therapeutic effects. Previous section has shown that HELNs can induce ferroptosis in prostate cancer cells by downregulating GPX4 expression. Therefore, we selected siRNA as the drug to be loaded into HELNs to synergistically induce ferroptosis. Three siGPX4 sequences (si1, si2, si3) were tested; si2 most effectively knocked down GPX4 mRNA (Fig. 7A) and protein (Fig. 7B) in 22Rv1 and LNCaP cells and was selected for loading. In subsequent experiments, siGPX4 refers to the siRNA of the si2 sequence. siGPX4 was loaded into HELNs via reversible electroporation to construct siGPX4-loaded HELNs, designated as siGPX4@HELNs. A si2 sequence with Cy3 fluorescence at the 5’ end was constructed. To quantify the loading efficiency achieved by reversible electroporation, we collected the supernatant after electroporation and centrifugation and measured its fluorescence. The fluorescence intensities of PBS alone, free siGPX4, the supernatant from HELNs co-incubated with siGPX4, and the supernatant after electroporation are presented in Figure S6A, B. The data show that electroporation achieved ~50% loading of siGPX4, whereas co-incubation reached ~22%. Fluorescence microscopy images display the siGPX4 (red) that has been delivered into the cells (Fig. 7C and Figure S6C). A small portion of free siGPX4 entered cells but failed to function effectively due to its susceptibility to degradation in the external environment (see subsequent results). In contrast, the combination of HELNs and siGPX4 showed higher fluorescence intensity, suggesting that some siGPX4 was encapsulated and protected by HELNs during culture. The strongest fluorescence was observed in the siGPX4@HELNs group, confirming that HELNs successfully and efficiently delivered siGPX4 into cells.

Fig. 7.

siGPX4@HELNs show good therapeutic effects on prostate cancer both in vitro and in vivo, with safety. (A) mRNA expression levels of GPX4 in different siRNA sequences treated 22Rv1 and LNCaP cells. (B) Protein expression levels of GPX4 in si2 sequence treated 22Rv1 and LNCaP cells. (C) The delivery efficiency of siGPX4 by HELNs in 22Rv1 cells. DAPI (blue): nuclei, cy3 (red): siGPX4. (D) Protein expression levels of GPX4 in 22Rv1 and LNCaP cells. (E) Flow cytometry detecting the intracellular Fe²⁺ levels of 22Rv1 cells. (F) Quantification of Fe²⁺ levels in 22Rv1 and LNCaP cells. (G) Flow cytometry detecting the intracellular peroxidized lipid levels of 22Rv1 cells. (H) Quantification of peroxidized lipid levels in 22Rv1 and LNCaP cells by MFI. (I) Cytotoxic effects of siGPX4@HELNs on 22Rv1 and LNCaP cells detected by CCK-8 assay. (J) Schematic diagram of animal experiment. (K) Photo of tumors at the end of experiments (n = 5). (L) Tumor growth profiles in tumor-bearing mice. (M) Tumor volume at the end of experiments. (N) Tumor weight at the end of experiments. (O) Body weight changes of tumor bearing mice. *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001

To verify the therapeutic efficacy of siGPX4@HELNs, further experiments were conducted. Results of western blot showed siGPX4@HELNs more effectively reduced GPX4 protein levels than HELNs alone or the mixture of siGPX4 and HELNs (Fig. 7D). Both the HELNs group and the mixture group also significantly reduced GPX4 protein expression compared to the control group. In contrast, free siGPX4 alone had little impact on GPX4 protein expression. This indicates that siGPX4@HELNs is an effective drug for reducing GPX4 protein expression in prostate cancer cells and is superior to HELNs. Subsequently, the study assessed the ferroptosis-inducing capability of siGPX4@HELNs. siGPX4@HELNs induced more pronounced Fe²⁺ accumulation (Fig. 7E, F) and lipid peroxidation (Fig. 7G, H) in 22Rv1 and LNCaP cells (Figure S6D, E) than other treatments. The CCK-8 assays confirmed siGPX4@HELNs resulted in the lowest cell viability (Fig. 7I). In summary, the strategy of using HELNs as a drug carrier to deliver siGPX4 for treating prostate cancer was successful. Compared to HELNs, the constructed siGPX4@HELNs induced more pronounced ferroptosis in prostate cancer cells and significantly decreased cell viability.

The therapeutic efficacy and safety of siGPX4@HELNs in vivo were also evaluated, with nude mice inoculated with 22Rv1 cells divided into 5 groups for intervention (Fig. 7J). The results showed that the tumor volume in the siGPX4@HELNs group was significantly smaller than that in the other four groups (Fig. 7K-M). The tumor weight differences were consistent with the tumor volume differences (Fig. 7N). Monitoring curves of mouse body weight indicated that during the treatment period, there were no significant changes in body weight among the groups (Fig. 7O). Immunohistochemical detection confirmed the lowest GPX4 expression in the siGPX4@HELNs group (Figure S6F). The safety of siGPX4@HELNs was also examined. H&E staining showed that there were no significant morphological and structural changes in the main organs of mice after various interventions (Figure S6G). Biochemical blood tests showed no significant differences in liver or kidney function across the groups after the interventions (Figure S6H).

Discussion

In recent years, research on PELNs for the treatment of human diseases has been on the rise. Various types of plants have been explored for PELNs extraction and therapeutic applications, with exosome-like nanoparticles derived from Chinese herbal medicine being particularly noteworthy. Nanoparticles derived from well-known traditional Chinese herbal medicines, such as ginseng, camellia, and garlic, have shown great promise in treating various diseases, particularly in cancer therapy [33–36]. This study innovatively extracted exosome-like nanoparticles from Huangqi (Astragalus membranaceus) and systematically elucidated the mechanisms of HELNs against prostate cancer through integrated multi-omics analysis. This study provides new insights and ideas for the future application of PELNs in cancer treatment.

HELNs isolated from Astragalus membranaceus harbor a repertoire of anticancer compounds. Compared with single-component extracts, these nanoparticles deliver multiple bioactives simultaneously, thereby evoking synergistic antitumor effects. Their nanoscale bilayer architecture prolongs systemic circulation, shields the encapsulated cargo from enzymatic degradation, and favors prolonged retention within the tumor microenvironment. Moreover, relative to crude Astragalus extracts, HELNs exhibit significantly prolonged retention within the tumor microenvironment and substantially mitigate the risk of hepatorenal toxicity. The size of HELNs is around 110 nm, which is close to the size of most reported PELNs. The majority of PELNs have a spherical bilayer membrane structure, although there are also reports of oval and cup-shaped PELNs, which may be related to the extraction methods and are considered to be artificially induced deformations [37]. HELNs have a conventional spherical shape. Lipids form the fundamental structure of PELNs, and this lipid bilayer serves as a barrier for the contents of PELNs, protecting them from relatively harsh external environments [38]. Liquid chromatography and mass spectrometry analyses have revealed the main lipid components of HELNs. HELNs contain lipid components commonly found in PELNs, such as PA, PE, and PC, but it has a high content of Cer. Cer is a well-established pro-apoptotic sphingolipid, and its abundance in HELNs likely contributes to the potent killing of prostate cancer cells. When detecting the uptake rates of HELNs by different cells, we found that the four uptake rates of HELNs were quite different. This heterogeneity likely reflects both intrinsic cellular properties and the specific lipid composition of HELNs. Each cell line originates from a different patient, so baseline proliferation, drug resistance, and endocytic capacity vary. In addition, the lipid profile of the vesicles themselves influences uptake efficiency; particles whose lipid composition more closely matches that of the target-cell membrane are generally internalized more readily.

Macrophages are immune cells involved in both innate and adaptive immunity, possessing the ability to destroy and phagocytose tumor cells. However, some macrophages in TME are influenced by the environment and various signaling molecules, polarizing into M2 macrophages. M2 macrophages facilitate tumor immune evasion, metastasis, and angiogenesis, and are thus also known as tumor-associated macrophages (TAMs) [39]. When macrophages polarize to the M2, they highly express two characteristic proteins—CD206 and ARG1. Arginase 1 (ARG1) can modulate arginine metabolism, thereby affecting antitumor immunity and promoting tumor cell immune evasion [40, 41]. Therefore, reversing the polarization state of M2 macrophages and reducing their expression of ARG1 has become a highly effective antitumor strategy. Starting from single-cell sequencing and verified through in vitro and in vivo experiments, this study has clarified that HELNs can reverse the polarization of M2 macrophages, re-polarizing them into M1 macrophages. Regarding PELNs targeting macrophages for tumor therapy, there have been several previous reports, with the most extensively studied being ginseng PELNs. Ginseng PELNs can reverse the M2 polarization of macrophages and exert effects against melanoma and glioma [42, 43]. Lv et al. further investigated the mechanisms by which ginseng PELNs improve macrophage polarization and treat colorectal cancer [13]. They found that ginseng PELNs can reprogram TAMs into M1-polarized macrophages by reducing the expression of ARG1 in TAMs. HELNs appear to be more advantageous than other PELNs in regulating the local tumor immune microenvironment. This is attributed to their induction of ferroptosis in PCa cells, leading to the release of abundant tumor antigens and the subsequent activation of immune cells. We propose that this synergistic model of ferroptosis and immunomodulation makes HELNs a highly promising anti-tumor agent. Nonetheless, it is important to note that while our study demonstrates the regulatory role of HELNs on macrophages, the specific mechanisms behind this effect remain to be fully elucidated in future studies.

Ferroptosis is a form of cell death dependent on reactive oxygen species (ROS), with its two main characteristics being iron overload and lipid peroxidation [29]. Iron exists in two oxidation states: ferrous (Fe²⁺) and ferric (Fe³⁺), and the redox cycling of iron may influence cellular sensitivity to ferroptosis [44]. Intracellular ferrous ions can generate ROS through the Fenton reaction, causing oxidative damage to cells, and can also affect ferroptosis by influencing the activity of redox enzymes such as lipoxygenase [30]. An accumulation of evidence shown that ferroptosis is a potential anticancer mechanism, especially since some drug-resistant or metastatic cancer cells are sensitive to ferroptosis [45, 46]. Ghoochani et al. demonstrated that advanced prostate cancer is sensitive to ferroptosis, and inducing ferroptosis in cancer cells is a very promising therapeutic approach for treating late-stage prostate cancer [47]. Accumulating evidence also indicates that ferroptosis is a crucial factor in suppressing prostate cancer growth, and inducing ferroptosis in prostate cancer cells can enhance their sensitivity to enzalutamide [48–50]. Against this backdrop, targeting ferroptosis for prostate cancer represents a novel and highly promising direction for drug development. In this study, we found that HELNs can induce the accumulation of ferrous ions and lipid peroxidation in prostate cancer cells in a concentration-dependent manner, indicating that HELNs can exert anti-prostate cancer effects by inducing ferroptosis. Given that ferroptosis is a form of cell death mediated by lipid peroxidation, the antitumor effect of HELNs against prostate cancer may not be solely attributable to their modulation of ferroptosis-related proteins. It is also plausible that HELNs provide a rich source of lipids that contribute to this process. Previous studies have reported that the direct tumor-killing effects of PELNs involve inducing cancer cell apoptosis and cycle arrest [17, 51]. Compared to these PELNs, HELNs offer a greater advantage in treating prostate cancer by inducing ferroptosis, which can be attributed to the specific sensitivity of prostate cancer to this form of cell death. This finding provides a rationale for the subsequent development of combination therapies involving HELNs and other ferroptosis-inducing agents.

There are two strategies for loading drugs into PELNs: active and passive loading. Active loading involves using methods such as ultrasound, electroporation, freeze - thawing, and membrane extrusion to temporarily disrupt the bilayer lipid structure of PELNs and allow drug entry. Passive loading typically refers to co - incubation, where drugs diffuse into PELNs due to their physicochemical properties [52, 53]. There are reports of PELNs being used as carriers to deliver siRNA. For example, grapefruit PELNs loaded with siRNA successfully performed gene knockout in HaCaT cells. The researchers used microfluidic technology to achieve a loading efficiency of 11% [54]. Zhang et al. used membrane extrusion to load siRNA targeting CD98 into ginger PELNs to knock down CD98 in intestinal tissues [55]. Electroporation is a common loading method, typically used for small molecules like siRNA, with loading efficiency adjustable by the ratio of PELNs to drug and the applied voltage [56]. Our study confirmed that HELNs can be used as drug carriers to deliver siRNA. The constructed siGPX4@HELNs significantly reduced GPX4 expression in prostate cancer cells, with much better effects than HELNs alone. siGPX4@HELNs showed good efficacy against prostate cancer in vivo, with no observed organ damage, indicating that they are a promising drug for treating prostate cancer.

PCa is characterized by its substantial heterogeneity, driving continuous innovation in novel therapeutic agents. Recent years have witnessed the emergence of numerous small-molecule and nanomaterial-based drugs targeting PCa [57, 58], alongside growing research into TCM for inducing ferroptosis [59]. HELNs represent an innovative strategy that merges the beneficial properties of TCM with the advantages of a nano-delivery system, establishing them as a promising nanocarrier for PCa therapy. This positions HELNs as a potential synergistic partner for existing clinical regimens. For instance, sensitizing PCa cells to ferroptosis can enhance the therapeutic efficacy of agents like docetaxel [60]. Looking ahead, HELNs could be utilized in combination with chemotherapy drugs, such as docetaxel, to treat prostate cancer, potentially allowing for dose reduction—a key prospect for their clinical translation. This study provides a comparatively comprehensive investigation into the application of HELNs for prostate cancer therapy, offering valuable insights for the advancement of this field. However, it is important to acknowledge that there are still several limitations in this study, and further in-depth exploration is needed in the future: (1) Due to current limitations in sequencing technology, the specific bioactive constituents within HELNs responsible for the observed anti-tumor effects have not been fully elucidated. Future studies are essential to precisely identify these active components. (2) The long-term safety profile of HELNs requires further investigation and evaluation. (3) The precise mechanism by which HELNs induce ferroptosis in tumor cells merits in-depth exploration. A clearer understanding of this pathway could pave the way for designing optimal synergistic drug combinations to enhance therapeutic efficacy. (4) Similar to other PELNs, current isolation techniques may not fully exclude the co-isolation of minor plant-derived impurities. The presence of these impurities represents a significant challenge that must be addressed for successful clinical translation of HELNs.

Conclusions

This study systematically explored the therapeutic effects and mechanisms of HELNs in treating prostate cancer through an integrated multi-omics approach. HELNs demonstrated robust cytotoxic effects against prostate cancer both in vitro and in vivo. In terms of indirect effects, HELNs reversed the polarization of M2 macrophages and increased the abundance of antitumor probiotics in the gut. Regarding direct effects, HELNs reduced the expression of GPX4 protein in prostate cancer cells, thereby inducing ferroptosis. Finally, HELNs served as a carrier to deliver siGPX4, achieving better therapeutic outcomes while maintaining safety.

Author contributions

Ye An, Jin-Zhou Xu and Gui-Chen Ye: Investigation; Visualization; Writing - original draft. Jian-Xuan Sun, Jia-Cheng Xiang, Chen Gong, Si-Han Zhang, Lin-Tao Miao, Si-Yang Ma: Writing - review & editing. Ming-Xia Ding: Supervision; Writing - review & editing. Shao-Gang Wang and Qi-Dong Xia: Supervision; Conceptualization; Funding acquisition.

Funding

This research was supported by Tongji Hospital Medical Innovation and Transformation Incubation Project (2022ZHFY02), and National Natural Science Foundation of China (NSFC, 82370770).

Data availability

The raw data used to support the conclusions of this article will be made available by the corresponding authors. The raw data of single cell sequencing, bulk-RNA sequencing and 16 S rDNA sequencing was uploaded in SRA database with the identifier: PRJNA1270733, PRJNA1270779 and PRJNA1271230. The data will be made public after the publication of the article.

Declarations

Ethics approval and consent to participate

The animal experiments in this study were approved by the Experimental Animal Welfare and Ethics Committee of Tongji Hospital Affiliated to Tongji Medical College, Huazhong University of Science and Technology, approval No.: TJH-202411033.

Competing interests

The authors declare no competing interests.