Dual-functional nanovesicles simultaneously inhibit stromal fibrosis and angiogenesis to suppress cholangiocarcinoma progression Authors: Lijing Zhang, Xixi Duan, Qiying Shi, Xiaohan Yao, Qiang Chen, Jiajia Wan, Fazhan Wang, Chen Ni, Yongjuan Li, Ming Wang, Yuqiao Sheng, Wenfu Zheng, Juan Liu, Tianjiao Ji, Zhihai Qin Published: 2025 ABSTRACT Cholangiocarcinoma (CCA), the second most prevalent primary hepatic malignancy, demonstrates resistance to antiangiogenic therapy due possibly to the dynamic interaction between cancer-associated fibroblast (CAF)-mediated extracellular matrix (ECM) remodeling and angiogenesis. This study shows that anti-VEGFR2 therapy activates CAF, inducing excessive ECM deposition and forming a physical barrier that diminishes the effectiveness of antiangiogenic therapy in CCA. Based on the finding, we rationally engineered vascular endothelial cell-derived nanovesicles that inherit the angiogenic factor receptor that competitively bind and neutralize pro-angiogenic ligands to diminish their bioeffects. These nanovesicles also retain high levels of integrin αvβ3 and specifically carry peptide B7-33 (an inducer of fibroblasts quiescence) modified by the cRGD peptide, thereby developing dual-functional nanovesicles (B7-33-SNPs). The study revealed that B7-33-SNPs synergistically disrupted the CAF-angiogenesis crosstalk, effectively reducing microvessel density and fibrotic deposition in subcutaneous xenograft CCA models. This combinatorial strategy achieved a 67.7 ± 17.6% tumor growth inhibition rate through simultaneous targeting of stromal desmoplasia and vascular niches (p < 0.001 vs. PBS group). This tumor microenvironment reshaping strategy, which concurrently inhibits CAF activation and angiogenesis, offers a promising alternative for suppressing malignant CCA.Graphical abstract Supplementary InformationThe online version contains supplementary material available at 10.1186/s12951-025-03833-w. Introduction Cholangiocarcinoma (CCA), the second most prevalent primary liver malignancy, constitutes approximately 3% of gastrointestinal neoplasms and demonstrates a rising global incidence and mortality [1, 2]. Although surgical resection remains the only curative option, more than 70% of patients are diagnosed with advanced, inoperable stage [3, 4]. First-line gemcitabine/cisplatin chemotherapy offers limited clinical benefit due to the rapid onset of therapeutic resistance, highlighting the urgent need for novel therapeutic strategies [5–7]. Emerging evidence underscores the critical role of the tumor microenvironment in CCA progression correlating with worse overall survival [8–10], particularly antiangiogenic therapy (e.g., bevacizumab; Ramucirumab) and targeting cancer associated fibroblast (CAF) (e.g., Rilotumumab; Plerixafor) have revolutionized cancer therapy across various malignancies, but no monotherapy has shown significant efficacy in improving CCA survival in preclinical trials to date [11–13]. Specifically, angiogenesis driven by VEGFA and ANG-2, which are key prognostic factors promoting vascularization and tumor progression, providing oxygen and nutrients to rapidly proliferating cells [8, 14]. Although angiogenesis in CCA can be inhibited by targeting VEGFA or ANG2 signaling pathways with monoclonal antibodies or multi-targeted tyrosine kinase inhibitors [12, 15, 16], therapeutic efficacy is limited by resistance and suboptimal outcomes [17, 18]. Recent preclinical studies have demonstrated that antiangiogenic therapy elevates ECM expression, which modulated tumor angiogenesis and therapeutic response in primary tumors, as a consequence of exacerbated hypoxia [19, 20]. The CCA microenvironment is defined by extensive fibrosis and dense extracellular matrix (ECM) deposition driven by cancer associated fibroblast (CAF), which promote tumor progression, and lead to poor clinical outcomes as well as treatment resistance [21–25]. Indeed, previous studies have reported that CAF secreted cytokines, such as interleukin (IL)−6 and C-X-C motif chemokine ligand 12 (CXCL12), along with ECM proteins, which enhanced tumor cell growth and aggressiveness [26–29]. Moreover, a high abundance of α-Smooth Muscle Actin (α-SMA)-positive CAF has been linked to accelerated tumor growth, enhanced migration and invasiveness, and poor survival outcomes in CCA patients [30–32]. Additionally, ECM proteins secreted by CAF not only provide structural support but also form a physical barrier to therapeutic penetration, thereby contributing to resistance against antiangiogenic agents [33–35] Furthermore, recent studies have identified a critical paracrine loop in which CAF upregulate stromal cell-derived factor 1 (SDF-1/CXCL12) and vascular endothelial growth factor A (VEGFA) expression, establishing a pro-angiogenic niche that diminished the efficacy of anti-VEGFA therapy [36, 37]. Despite these advances, the therapeutic potential of concurrently targeting CAF activation and angiogenesis inhibition remains largely unexplored in preclinical CCA models. Relaxin is a well-recognized potent antifibrotic hormone, which inhibits fibroblast activation by binding to its primary receptor (relaxin family peptide receptor 1, RXFP1) via targeted delivery of relaxin-encoding plasmids, proteins, or peptides [38–40]. We hypothesize that integrating relaxin-based antifibrotic therapy with anti-angiogenic strategies within a single nanosystem may yield synergistic antitumor efficacy. Herein, we report B7-33, a rationally designed relaxin peptide analog conjugated with cyclic arginine-glycine-aspartic acid (cRGD), can assemble into mouse endothelial cell line sEND.1-derived nanovesicles (SNPs) through specific interaction with integrin αvβ3 mediated by the ligand–receptor interaction [40], thereby developing dual-functional nanovesicles (B7-33-SNPs). B7-33-SNPs were derived from the mouse endothelial cell line sEND.1, inheriting critical receptors such as Tie2 and VEGFR2. These receptors enable the nanoparticles to sequester VEGFA and ANG2 ligands, thereby disrupting pro-angiogenic signaling in host endothelial cells. Notably, the sEND.1 cell membrane retains high level integrin αvβ3 expression, facilitating the specific delivery of the cRGD-conjugated B7-33 peptide to targets RXFP1 on activated CAF, thereby inhibiting TGF-β-mediated pSMAD2/3 signaling and reversing CAF’ activation and ECM remodeling. This nanosystem simultaneously disrupts CAF-mediated fibrosis by reversing CAF activation through B7-33 and inhibits angiogenesis via receptor-mediated sequestration of angiogenic factors, demonstrating a dual-functionality. By integrating these complementary therapeutic approaches, our biomimetic platform offers a novel strategy to synergistically inhibit CCA progression through simultaneous targeting of tumor stroma fibrosis and angiogenesis (Scheme 1). Scheme 1Schematic representation of the dual-targeting B7-33-SNPs biomimetic nanoplatform designed to synergistically inhibit CCA progression through simultaneous suppression of CAF activation and angiogenesis. B7-33-SNPs inherit endothelial cell membrane-derived receptors (e.g., VEGFR2, Tie2), which competitively sequester angiogenic ligands (VEGFA, ANG2), thereby inhibiting angiogenesis. Simultaneously, B7-33-SNPs interact with CAF-surface relaxin family peptide receptor 1(RXFP1), blocking TGF-β-induced SMAD2/3 phosphorylation and consequently preventing both CAF activation and ECM deposition. This combinatorial strategy offers a novel approach to overcoming resistance to antiangiogenic therapy in malignant CCA. Targeted inhibition of CAF activation could enhance the efficacy of antiangiogenic therapy and significantly suppress CCA growth Schematic representation of the dual-targeting B7-33-SNPs biomimetic nanoplatform designed to synergistically inhibit CCA progression through simultaneous suppression of CAF activation and angiogenesis. B7-33-SNPs inherit endothelial cell membrane-derived receptors (e.g., VEGFR2, Tie2), which competitively sequester angiogenic ligands (VEGFA, ANG2), thereby inhibiting angiogenesis. Simultaneously, B7-33-SNPs interact with CAF-surface relaxin family peptide receptor 1(RXFP1), blocking TGF-β-induced SMAD2/3 phosphorylation and consequently preventing both CAF activation and ECM deposition. This combinatorial strategy offers a novel approach to overcoming resistance to antiangiogenic therapy in malignant CCA. Targeted inhibition of CAF activation could enhance the efficacy of antiangiogenic therapy and significantly suppress CCA growth Materials and methods Materials Sucrose, Ethylenediaminetetraacetic acid (EDTA), Hepes-NaOH were purchased from Solorbio (Beijing, China). 5-Chloromethylfluorescein Diacetate (CMFDA) was obtained from Yeasen Biotechnology (Shanghai, China). Dulbecco’s Modified Eagle Medium (DMEM) and phosphate-buffered saline (PBS) were obtained from HyClone (Logan City, USA). Matrigel matrix, Collagen type I and Matrigel were purchased from Corning (Chicago, USA). CXCL12, Recombinant Mouse VEGFA/VEGF164 Protein, Mouse VEGFA ELISA Kit and Recombinant Mouse ANG-2 Protein were obtained from ABclonal (Boston, USA). Mouse ANG-2 ELISA Kit was purchased from Beijing Biaolaibo Technology (Beijing, China). Primary antibodies against VEGFR2 (Rabbit anti-mouse) and CD31 (Rat anti-mouse) were obtained from Cell Signaling Technology (Boston, USA). GAPDH antibody, Fibronectin antibody, and Collagen I antibody were obtained from Sanying (Wuhan, China). A VEGFR2 - neutralizing antibody was acquired from BioXCell (New Hampshire, USA). Mouse TGF-β Proteins and mouse RXFP1 antibody were purchased from R&D Systems (Minneapolis, USA). Platelet-derived growth factor receptor beta (PDGFR-β) antibody was purchased from Thermo Fisher Scientific (Waltham, USA). Sodium Potassium ATPase antibody, Tie2 antibody and α-SMA antibody and integrin αvβ3 antibody were acquired from Abcam (Cambridge, UK). Cell lines The immortalized mouse embryonic fibroblast (iMEF) cell line was generously provided by Prof. Lin Xi (Tsinghua University, Beijing, China). Human cholangiocarcinoma (HUCCT1) and murine endothelial (sEND.1) cell lines were cultured in DMEM (Cat. #SH30022.01B; HyClone) supplemented with 10% fetal bovine serum (FBS) (Cat. #S829012029; Invigentech), penicillin (100 U/mL), and streptomycin (100 µg/mL) (Cat. # sv30010; HyClone), with media refreshed every 2–3 days. Cells were cultured under standard conditions (37 °C, 5% CO2, 95% humidity). Mice Six-week-old male BALB/c nude mice were obtained from Vital River Laboratory Animal Technology Co., Ltd (Beijing, China). All animal procedures were approved by the Institutional Animal Care and Use Committee (IACUC) and Research Ethics Committee of the First Affiliated Hospital of Zhengzhou University (approval number: 2024-KY-1113-001). Antiangiogenic treatments Nude mice bearing the HUCCT1 implanted tumors were divided into treatment groups receiving: (1) PBS, (2) anti-VEGFA monoclonal antibody B20.4-1.1.1.1 (Genentech; 5 mg/kg, i.p., twice weekly) or (3) SNPs (12 mg/kg, i.v., every 48 h). Treatment durations were set at 1.5 weeks (3 doses for anti-VEGFA/5 doses for SNPs) or 3 weeks (6 doses for anti-VEGFA/10 doses for SNPs), with 3–5 mice per group euthanized at experimental endpoints. Excised tumors were fixed in 4% paraformaldehyde (4 °C, overnight), paraffin-embedded, and sectioned (5 μm thickness). Immunohistochemical staining was performed using antibodies against CD31 (Cat. #550274; BD Biosciences), α-SMA (Cat. #19245S; CST), Fibronectin (Cat. #15613-1-AP; Proteintech), and collagen type I (Cat. #14695-1-AP; Proteintech). Positive protein expression areas were quantified in five randomly selected fields per sample using ImageJ, with uniform thresholding applied across all analyses. Preparation of nanoparticle The cRGD-B7-33, and cRGD-C7-33 were synthesized via custom peptide synthesis services provided by Anhui Guoping Pharmaceutical Co. Ltd. as described previously [40]. Purity (>95%) and identity were confirmed by HPLC and ESI-MS. The sEND.1 membrane derived nanoparticles (SNPs) was prepared by gradient centrifugation and polycarbonate membranes extrusion in the previous study. Then cRGD-B7-33 was mixed with SNPs to obtain B7-33-SNPs (4 °C, overnight). The fluorescence labeled B7-33-SNPs were also prepared according to the previous experimental methods [40]. For red blood cells membrane-derived nanoparticles (RNPs), whole blood was subjected to sequential centrifugation steps to isolate red blood cells (RBCs). After PBS washing, RBCs underwent hypotonic lysis (0.25× PBS, 4 °C), followed by extrusion through 200 nm polycarbonate membranes (Avanti Mini Extruder, Avanti Polar Lipids). Characterization of B7-33-SNPs Nanoparticles were negatively stained with 2% uranyl acetate and analyzed using a TECNAI G2 F20 transmission electron microscope. Hydrodynamic diameter and zeta potential were measured via dynamic light scattering. The preserved membrane proteins of B7-33-SNPs were examined by SDS-PAGE with Coomassie brilliant blue staining and Western blot. Protein concentrations, adjusted to 1 mg/mL via BCA assay, were resolved on 10% SDS-PAGE gels. The resulting gel samples were incubated with primary antibodies against Tie2 (Cat. #ab21915; Abcam), VEGFR2 (Cat. #9698S; CST), Integrin αvβ3 (Cat. #ab179475 and Cat. #ab119992; Abcam), and Na+/K + ATPase (Cat. #ab76020; Abcam), followed by enhanced chemiluminescence detection. For further characterization of the B7-33-SNPs, immunogold labeling was performed. B7-33-SNPs suspensions (10 µL) were adsorbed onto carbon-coated grids (20 min, RT), washed sequentially with PBS and 50 mM glycine. After blocking with 1% BSA (15 min), the grids were incubated with anti-Tie2 antibody and anti-VEGFR2 antibody (1 h, RT), followed by anti-rabbit IgG-gold conjugates (5 nm particles, 1 h) (Cat. #ab41498; Abcam). Subsequently, the grid was washed with PBS and fixed with 1% glutaraldehyde (5 min) and finally negatively stained with 2% uranyl acetate (5 min) and visualized by transmission electron microscopy. B7-33-SNPs or B7-33-RNPs (0–8 mg/mL) were incubated with recombinant murine VEGFA (Cat. #RP01060; ABclonal) and ANG2 (Cat. #RP01744; ABclonal) (10 ng/mL, 37 °C, 2 h). Unbound cytokines were removed via centrifugal filtration (100 kDa cutoff), and the concentrations of the centrifugal concentrate were quantified using mouse VEGFA (Cat. #RK00028; ABclonal) and ANG2 (Cat. #ZN2511; Beijing Biaolaibo Technology) Elisa kit. Binding curves were generated via nonlinear regression (Graphpad Prism software). In Vitro antiangiogenic efficacy assessment A wound healing assay was performed as described previously [41]. sEND.1 cells (1.5 × 105 cells/well in 96-well plates) were scratched to create a linear wound using a sterile pipette tip. Cells were treated with B7-33-RNPs, C7-33-SNPs, or B7-33-SNPs (cRGD-B7-33: 20 nmol/L; SNPs or RNPs: 500 µg/mL) in VEGFA-supplemented medium (2 ng/mL, 1% FBS). Photographs were taken at 0 h and 24 h using phase-contrast microscopy. Transwell assays were conducted following a previously reported protocol [41]. sEND.1 cells (2 × 105 cells/well) were plated in Transwell upper chambers (8 μm pore size) using 1% FBS medium supplemented containing the nanotherapy at the aforementioned concentration. The lower chamber contained chemoattractant medium (1% FBS + 2 ng/mL VEGFA). After 24 h, migrated cells on the basal membrane were fixed (4% paraformaldehyde), stained with crystal violet, and counted (ImageJ). The tube formation assay was processed in accordance with the established method [41]. Matrigel (8–12 µg/mL, 120 µL/well) was polymerized in 24-well plates (37 °C, 30 min). sEND.1 cells (8 × 103 cells/well) were seeded onto the gel and treated with the nanotherapy at the aforementioned concentration in VEGFA-supplemented medium (2 ng/mL, 1% FBS). Tubular networks were imaged after 6 h using using phase-contrast microscopy. The amounts of junctions and meshes were counted with ImageJ software. B7-33-SNPs inhibit the activation of fibroblast Gel contraction assays The fibroblast cell line iMEF (1.5 × 105 cells) was embedded in 100 µL collagen type I (Cat. #354249; Corning)/Matrigel (Cat. #354230; Corning) matrices (4 mg/mL collagen, 2 mg/mL Matrigel) in 96-well plates. Following polymerization, cells were pretreated for 1 h with different B7-33 formulations at the aforementioned in 1% FBS medium, followed by 48 h stimulation with TGF-β (5 ng/mL; Cat. #7666-MB-005; R&D Systems). Gel contraction was monitored using stereomicroscopy, with relative diameters of wells and gels quantified via ImageJ to calculate contraction values. In vitro effects of fibroblast activation inhibition: iMEF cells (4 × 104 cells/well in 6-well plates) were pretreated for 1 h with different B7-33 formulations at the aforementioned in 1% FBS medium, followed by 48 h stimulation with TGF-β (5 ng/mL) [40]. Cells were lysed, and proteins were resolved by SDS-PAGE for immunoblotting with antibodies targeting α-SMA, fibronectin, collagen-1, and GAPDH. HRP-conjugated secondary antibodies (anti-rabbit or anti-goat IgG) were used for chemiluminescent detection. Band intensities were analyzed using ImageJ. In Vitro B7-33-SNPs inhibit fibroblast activation-mediated angiogenesis Transwell migration assay: In vitro cell migration assay was conducted using 8 μm pore-sized hanging cell culture inserts. sEND.1 cells were stained with CMFDA for 20 min and resuspended in 1% FBS medium and seeded at a density of 5 × 104 cells per well in transwell upper chambers. ‌The transwell chambers were then placed into 24-well plates with the lower chambers containing iMEF or TGF-β-stimulated iMEF pretreated with different treatments at the aforementioned in the lower chamber. After 12 h incubation (37 °C), CMFDA-labeled sEND.1 cells were fixed (4% PFA, 15 min) and non-migratory cells were removed with cotton swabs. Finally, migrated cells were quantified via fluorescence microscopy (492/517 nm) using ImageJ Pro. Five random high-power fields (× 200 magnification) were analyzed per sample to calculate cell migration indices. 3D Spheroid sprouting assay: The spheroid sprouting assay was performed by combining sEND.1 cells with iMEF (1:1 ratio) or TGF-β- stimulated iMEF pre-incubated‌ with different treatments at the aforementioned in methylcellulose-supplemented DMEM (0.24% w/v), cultured as 20 µl hanging drops (37 °C, overnight) [42]. Subsequent embedding in Matrigel induced sprouting over 24 h. Fixed spheroids (4% PFA) were imaged by confocal microscopy, with sprout lengths measured through ImageJ analysis. All assays included five technical replicates across three independent experiments. RNA extraction and quantitative real-time polymerase chain reaction (qPCR): Total RNA was extracted with TRIzol Reagent (9108, Takara) following the manufacturer’s protocol. Reverse transcription of 1 µg RNA was performed using PrimeScript™ RT Kit (RR047A, Takara). Quantitative PCR (qPCR) was conducted with SYBR Premix Ex Taq II (RR820A, Takara) on an Agilent Mx3005P system, with triplicate measurements for each sample. The primers used for the qPCR as follows: Acta2 forward: TGCTGACAGAGGCACCACTGAA, reverse: CAGTTGTACGTCCAGAGGCATAG; Fn1 forward: CCCTATCTCTGATACCGTTGTCC, reverse: TGCCGCAACTACTGTGATTCGG; Col1a1 forward: CCTCAGGGTATTGCTGGACAA, reverse: CAGAAGGACCTTGTTTGCCAGG; Col1a2 forward: TTCTGTGGGTCCTGCTGGGAAA, reverse: TTGTCACCTCGGATGCCTTGAG; Vegfa forward: GGAGATCCTTCGAGGAGCACTT, reverse: GGCGATTTAGCAGCAGATATAAGAA; Cxcl12 forward: TGCATCAGTGACGGTAAACCA, reverse: TTCTTCAGCCGTGCAACAATC; Tgfb1 forward: TGATACGCCTGAGTGGCTGTCT, reverse: CACAAGAGCAGTGAGCGCTGAA; Hgf forward: ATGTGGGGGACCAAACTTCTG, reverse: GGATGGCGACATGAAGCAG. Immunofluorescence (IF) staining of human CCA tissue CCA tissues and adjacent paracancerous tissues from patients were sourced from Shanghai Xinchao Biological Technology Co., LTD (Shanghai, China). All experiments involving human specimens were approved by the Ethical Committee of Medical Research, Shanghai Xinchao Biological Technology (No. SHYJS-BC-2310001). The paraffin-embedded sections were deparaffinized in xylenes and rehydrated in an ethanol gradient. Antigen retrieval was performed using sodium citrate buffer, and non-specific binding was blocked with 10% goat serum. Mouse RXFP1 antibody (Cat. #MAB8898; R&D Systems), CD31 mouse mAb (Cat. #3528; CST) and α-SMA (Cat. #ab5694; Abcam) were incubated overnight at 4 °C. Subsequently, fluorescence-conjugated secondary antibodies (Cat. #A31573 and Cat. #A21428; Invitrogen and Cat. #A-11034; Thermo Fisher Scientific) were added at room temperature for 1 h. Following this, nuclear staining was performed using DAPI, and the fluorescence signals were examined under a microscope. Quantitative assessment of positively stained regions was conducted using ImageJ software. In vivo biodistribution of B7-33-SNPs The biodistribution of B7-33-SNPs in CCA model was investigated. The xenograft tumor model was generated by subcutaneous injection of 5 × 106 HUCCT1 cells in 1:1 mixture of PBS/Matrigel (100 µL) into the flank region of the BALB/c nude mouse. Fluorescent dye Cy5.5-tagged formulations were administered via intravenous injection to the mouse. The dosage of Cy5.5 was 1.2 mg/kg. After 24 h, tumors were excised for IVIS spectrum imaging analysis. For histological evaluation, OCT-embedded tumor tissues were cryosectioned (7 μm) and immunostained using primary antibodies against α-SMA (CAF marker) and CD31 (endothelial cell marker) at 1:200 dilution (4 °C, overnight), followed by appropriate Alexa Fluor-conjugated secondary antibodies. Anti-tumor therapeutic study Nude mice bearing the HUCCT1 implanted tumors received an intravenous injection of PBS, cRGD-B7-33, cRGD-B7-33 + RNPs, SNPs, C7-33-SNPs, or B7-33-SNPs (cRGD-B7-33: 6.7 nmol/kg; SNPs: 12 mg/kg) (n = 7 in each group). The mice were euthanized at the experimental endpoint and tumor tissue were fixed (4% PFA, 4 °C), paraffin-embedded, and sectioned (5 μm) for histopathological evaluation. Tumor sections were obtained for H&E staining and immunohistochemical (IHC) staining was performed using an anti-α-SMA rabbit monoclonal antibody (Cat. #ab5694; Abcam), an anti- collagen I rabbit polyclonal antibody (Cat. #14695-1-AP; Proteintech), an anti- fibronectin rabbit polyclonal antibody (Cat. #15613-1-AP; Proteintech), an anti- PDGFR-β rat monoclonal antibody (Cat. #14–1402-82; Thermo Fisher Scientific), an anti- CD31 mouse monoclonal antibody (Cat. #550274; BD Biosciences) and collagen (Masson trichromatic staining; Cat. #G1006-20ML; Servicebio) to determine the levels of protein expression. Protein-positive areas were quantitatively analyzed in six random visual fields per sample using ImageJ, with results expressed as the percentage of the total visual field. Evaluation of SNPs-Cy5.5 penetration in tumors Nude mice bearing the HUCCT1 implanted tumors received a 6.7 nmol/kg intravenous dose of the cRGD-B7-33 through the tail vein for two weeks, receiving a total of eight injections. Subsequently, these mice received an intravenous injection of Cy5.5- loaded SNPs at a dose of 1.2 mg/kg through the tail vein. One hour post-injection, the mice were imaged using the IVIS spectrum imaging system. Then OCT-embedded tumors tissues were cryosectioned (6 μm), fixed (4% PFA), and counterstained with DAPI for confocal microscopy to evaluate the distribution of nanoparticles. In Vivo assessment of toxicity of B7-33-SNPs Healthy BALB/c nude mice received an intravenous injection of PBS, cRGD-B7-33, cRGD-B7-33 + RNPs, SNPs, C7-33-SNPs, or B7-33-SNPs at the following dosages: cRGD-B7-33 (6.7 nmol/kg), SNPs (12 mg/kg). Treatments were administered twice daily for a total of eight doses. Following the treatment period, major organs were excised, processed for hematoxylin and eosin (H&E), and evaluated for toxicity-associated pathological changes. Statistical analysis Quantitative data are presented as mean ± SEM. Two-group comparisons were analyzed by Student’s t-test (GraphPad Prism 10.0), while Pearson’s r evaluated linear correlations. All experiments were independently replicated at least 3 times, with p < 0.05 considered statistically significant. Results and discussion Anti-VEGFR2 therapy Induces CAF activation and ECM remodeling in CCA Previous studies have demonstrated that activated CAF secrete excessive ECM components to regulate tumor angiogenesis and resistance to antiangiogenic therapy [34, 43]. However, whether this process additionally establishes a physical barrier through ECM deposition specifically in CCA remains unclear. To investigate the therapeutic efficacy of mouse endothelial cell line sEND.1-derived nanovesicles (SNPs) and a VEGFR2 - neutralizing antibody (anti-VEGFR2) in murine CCA models, six-week-old mice were administered PBS (control), anti-VEGFR2, or SNPs, with treatment durations of either 1.5 or 3 weeks. Following 1.5 weeks of treatment, both anti-VEGFR2 and SNPs significantly reduced primary tumor volume by 37.1% (p < 0.05 vs. control) and 47.3% (p < 0.01 vs. control), respectively (Fig. 1A and B). However, tumor growth re-emerged in the 3-week treatment groups, indicating an acquired antiangiogenic resistance that may be attributed to both the activation of compensatory feedback pathways (e.g., Ang2 pathway induction upon VEGFA signaling blockade) and the upregulation of ECM, which not only modulates tumor angiogenesis but also acts as a physical barrier in primary tumors [12, 19, 44]. Immunofluorescence analysis confirmed that 1.5-week treatment with either agent markedly suppressed CD31-positive tumor vasculature (Fig. 1C and D, and S1), aligning with the observed antiangiogenic effects. Strikingly, vascular reconstruction in the 3-week treatment groups demonstrated rapid regrowth, suggesting adaptive resistance to antiangiogenic therapy. These findings demonstrate that SNPs exert potent antiangiogenic and anti-tumor effects during the initial therapeutic phase, followed by a reversion to near-baseline activity during the revascularization phase. Notably, tumors from mice treated for 3 weeks exhibited histological evidence of CAF activation, characterized by elevated α-SMA expression and enhanced collagen I and fibronectin deposition compared to PBS-treated controls (Fig. 1C and D, and S2-4). These observations are consistent with prior reports of ECM upregulation in antiangiogenic therapy-resistant tumors [19]. Collectively, these data demonstrate that anti-VEGFR2-targeted therapy induces CAF-mediated ECM remodeling, which may compromise drug penetration and contribute to therapeutic resistance in CCA. Thus, combinatorial strategies targeting both angiogenesis and CAF activation may represent a promising approach to improve therapeutic outcomes in CCA. Fig. 1Anti-VEGFR2 and endothelial cell membrane-derived nanoparticles (SNPs) induce activation of CAF and deposition of ECM in HUCCT1 subcutaneous xenografts. (A) Schematic diagram of the murine CCA model. Mice were administered PBS, anti-VEGFR2, or SNPs via the following regimens: anti-VEGFR2 (5 mg/kg, intraperitoneal injection, twice weekly); SNPs (12 mg/kg, intravenous injection, every 48 h) for 1.5 or 3 weeks. (B) Post-treatment tumor volumes in mice for each treatment group. (C) Immunofluorescence analysis of stromal remodeling in HUCCT1 tumors. Representative images show CD31, ECM deposition (α-SMA, fibronectin, and collagen I) staining in tumor sections from mice treated with PBS, anti-VEGFR2, or SNPs (n = 6). Nuclei were counterstained with DAPI (blue). Scale bars: 100 μm. (D) Quantification of normalized CD31 and ECM expression. Results are shown as mean ± SD (n = 6); t-test indicated ‌*p < 0.05, **p < 0.01 and ***‌p < 0.001 Anti-VEGFR2 and endothelial cell membrane-derived nanoparticles (SNPs) induce activation of CAF and deposition of ECM in HUCCT1 subcutaneous xenografts. (A) Schematic diagram of the murine CCA model. Mice were administered PBS, anti-VEGFR2, or SNPs via the following regimens: anti-VEGFR2 (5 mg/kg, intraperitoneal injection, twice weekly); SNPs (12 mg/kg, intravenous injection, every 48 h) for 1.5 or 3 weeks. (B) Post-treatment tumor volumes in mice for each treatment group. (C) Immunofluorescence analysis of stromal remodeling in HUCCT1 tumors. Representative images show CD31, ECM deposition (α-SMA, fibronectin, and collagen I) staining in tumor sections from mice treated with PBS, anti-VEGFR2, or SNPs (n = 6). Nuclei were counterstained with DAPI (blue). Scale bars: 100 μm. (D) Quantification of normalized CD31 and ECM expression. Results are shown as mean ± SD (n = 6); t-test indicated ‌*p < 0.05, **p < 0.01 and ***‌p < 0.001 The characterization of the dual-targeting nanotherapy B7-33-SNPs B7-33-SNPs were synthesized using a previously described method [40], wherein the cRGD-modified relaxin peptide analog (cRGD-B7-33) was encapsulated into membrane-derived vesicles from the murine endothelial cell line sEND.1 (SNPs). B7-33-RNPs were prepared by mixing red blood cell membrane-derived nanoparticles (RNPs) with cRGD-B7-33, and these RNPs serve as an ideal negative control due to their low immunogenicity, high biocompatibility, absence of specific targeting capability, and structural stability [45, 46]. Transmission electron microscopy (TEM) revealed spherical nanoparticle morphology (Fig. 2A). Dynamic light scattering (DLS) analysis confirmed a narrow size distribution with an average hydrodynamic diameter of 120.3 ± 2.1 nm (PDI = 0.25 ± 0.02) and a zeta potential of −5.8 ± 1.9 mV, with stability maintained over 7 days (Fig. 2B). Using fluorescence spectrophotometry as previously described [40], we quantified the peptide-loading efficiency of B7-33 onto SNPs. The average number of B7-33 molecules per nanoparticle was 15.0 ± 1.7, with a conjugation yield of 47.69 ± 0.04%. To validate membrane protein retention during nanoparticle fabrication, B7-33-SNPs were initially characterized by SDS-PAGE with Coomassie brilliant blue staining (Fig. 2C). The protein profile of B7-33-SNPs closely matched that of sEND.1 cells, confirming preservation of membrane-associated proteins. Subsequent Western blot analysis was employed to detect key receptors, including VEGFR2 and Tie2 which are critical proangiogenic factor receptors expressed on endothelial cells, were retained on B7-33-SNPs membranes (Fig. 2D). As a negative control, red blood cell membrane-camouflaged nanoparticles (B7-33-RNPs) exhibited negligible expression of these proangiogenic receptors. Integrin αvβ3 expression in sEND.1 cells was additionally confirmed via Western blot (Fig. 2D). Immunogold staining with target-specific antibodies visually confirmed surface localization of these proteins on B7-33-SNPs (Fig. 2E), thereby conferring binding capacity to proangiogenic factors such as VEGFA and ANG2, the key mediators of angiogenesis. B7-33-SNPs, but not B7-33-RNPs, demonstrated dose-dependent neutralization of all tested angiogenesis cytokines (Fig. 2F). Collectively, these results confirm the successful engineering of B7-33-SNPs and their potent cytokine-neutralizing capacity in vitro. Fig. 2Preparation and Characterization of B7-33-SNPs. (A) The morphology of the B7-33-RNPs and B7-33-SNPs was observed using transmission electron microscopy. Scale bars: 200 nm. (B) Stability analysis of B7-33-SNPs. Hydrodynamic diameter (red) and polydispersity index (PDI, blue) of B7-33-SNPs were monitored via DLS for 7 days at 37 °C (n = 3). (C) SDS-PAGE analysis of membrane proteins from sEND.1 cells, B7-33-SNPs, and B7-33-RNPs. (D) Western blot validation of endothelial-specific receptors VEGFR2, Tie2, and integrin αvβ3 in sEND.1 cells, B7-33-SNPs, and B7-33-RNPs. Na⁺/K⁺ ATPase served as a loading control. (E) Immunogold TEM imaging of B7-33-SNPs labeled with anti-VEGFR2 and anti-Tie2 antibodies conjugated to 5 nm gold nanoparticles (black dots). Samples were counterstained with 2% uranyl acetate. Scale bar: 100 nm. (F) Cytokine binding capacity of B7-33-SNPs and B7-33-RNPs assessed using ELISA (provided by BioRender.com). Dose-dependent neutralization of VEGFA and ANG2 is presented (n = 3). Results are shown as mean ± SD Preparation and Characterization of B7-33-SNPs. (A) The morphology of the B7-33-RNPs and B7-33-SNPs was observed using transmission electron microscopy. Scale bars: 200 nm. (B) Stability analysis of B7-33-SNPs. Hydrodynamic diameter (red) and polydispersity index (PDI, blue) of B7-33-SNPs were monitored via DLS for 7 days at 37 °C (n = 3). (C) SDS-PAGE analysis of membrane proteins from sEND.1 cells, B7-33-SNPs, and B7-33-RNPs. (D) Western blot validation of endothelial-specific receptors VEGFR2, Tie2, and integrin αvβ3 in sEND.1 cells, B7-33-SNPs, and B7-33-RNPs. Na⁺/K⁺ ATPase served as a loading control. (E) Immunogold TEM imaging of B7-33-SNPs labeled with anti-VEGFR2 and anti-Tie2 antibodies conjugated to 5 nm gold nanoparticles (black dots). Samples were counterstained with 2% uranyl acetate. Scale bar: 100 nm. (F) Cytokine binding capacity of B7-33-SNPs and B7-33-RNPs assessed using ELISA (provided by BioRender.com). Dose-dependent neutralization of VEGFA and ANG2 is presented (n = 3). Results are shown as mean ± SD B7-33-SNPs inhibit VEGFA-induced endothelial cell migration To assess the antiangiogenic potential of B7-33-SNPs, we conducted wound healing, transwell migration, and tube formation assays employing sEND.1 endothelial cells. Control nanoparticles (C7-33-SNPs) were prepared by conjugating cRGD-C7-33 (a scrambled B7-33 sequence) with SNPs. As illustrated in Fig. 3A, VEGFA-stimulated sEND.1 cells achieved complete scratch wound closure within 24 h. Treatment with B7-33-SNPs significantly inhibited migration, resulting in 59.5 ± 9.1% residual wound area (p < 0.01 vs. VEGFA control; Fig. 3B). Consistent with wound healing data, transwell migration assays demonstrated a 60.1 ± 8.1% reduction in VEGFA-driven cell invasion upon B7-33-SNPs treatment (Fig. 3C). In contrast, B7-33-RNPs exerted no significant inhibitory effect on migration (p > 0.05 vs. VEGFA control; Fig. 3D). Capillary-like tube formation on Matrigel was assessed 6 h post-treatment. VEGFA-stimulated cells developed dense tubular networks. B7-33-SNPs treatment reduced branch formation by 59.2 ± 9.6% (p < 0.01), with resultant tubes exhibiting fragmentation and reduced length (Fig. 3E). B7-33-RNPs exhibited no inhibitory activity, confirming the specificity of the observed effects (Fig. 3F). Collectively, these findings establish that B7-33-SNPs effectively inhibit VEGFA-mediated endothelial cell migration and tube formation, thereby validating their antiangiogenic efficacy. Fig. 3Antiangiogenic activity of B7-33-SNPs in vitro. (A) Representative wound healing assay images of VEGFA-activated sEND.1 cells treated with B7-33-SNPs or B7-33-RNPs. Dashed lines indicate scratch boundaries. Scale bar: 200 μm. (B) Normalized quantification of wound closure relative to VEGFA control group (n = 3). (C) Transwell invasion assay images showing sEND.1 cell penetration through Matrigel-coated membranes following 24 h treatment with different formulations. Scale bar: 200 μm. (D) Normalized quantification of migrated cells relative to VEGFA control group (n = 3). (E) Tube formation assay images on Matrigel matrices following 6 h of treatment with different formulations. Scale bar: 200 μm. (F) Quantitative analysis of total tube length, normalized to the VEGFA control group. (n = 3). Results shown as mean ± SD; t-test indicated ‌**p < 0.01 Antiangiogenic activity of B7-33-SNPs in vitro. (A) Representative wound healing assay images of VEGFA-activated sEND.1 cells treated with B7-33-SNPs or B7-33-RNPs. Dashed lines indicate scratch boundaries. Scale bar: 200 μm. (B) Normalized quantification of wound closure relative to VEGFA control group (n = 3). (C) Transwell invasion assay images showing sEND.1 cell penetration through Matrigel-coated membranes following 24 h treatment with different formulations. Scale bar: 200 μm. (D) Normalized quantification of migrated cells relative to VEGFA control group (n = 3). (E) Tube formation assay images on Matrigel matrices following 6 h of treatment with different formulations. Scale bar: 200 μm. (F) Quantitative analysis of total tube length, normalized to the VEGFA control group. (n = 3). Results shown as mean ± SD; t-test indicated ‌**p < 0.01 Concurrent targeting of CAF and angiogenesis disrupts stromal-vascular crosstalk To evaluate the functional impact of B7-33-SNPs on murine fibroblast cells (iMEF), TGF-β-stimulated gel contraction assays were conducted. B7-33-SNPs treatment significantly attenuated TGF-β-induced matrix contraction versus PBS-treated controls (p < 0.01), resulting in a 51.5 ± 9.5% reduction in contractile area (Fig. 4A). This inhibitory effect was statistically equivalent to free B7-33 treatment (43.9 ± 10.3% reduction), suggesting preserved antifibrotic activity of B7-33 post-nanoparticle conjugation (Fig. 4B). B7-33-SNPs strongly inhibited TGF-β-induced α-SMA, fibronectin, and collagen I levels, as evidenced by Western blot (Fig. 4C-D), consistent with prior findings [40]. Furthermore, we examined the effect of B7-33-SNPs on downstream TGF-β signaling by assessing the phosphorylation levels of SMAD2 and SMAD3 proteins. Western blot analysis revealed that B7-33-SNPs significantly attenuated TGF-β-induced iMEF activation by inhibiting the p-SMAD2/SMAD2 and p-SMAD3/SMAD3 ratios (Figure. S5A, B). These data demonstrate that B7-33-SNPs markedly reduce fibroblast activation by blocking TGF-β signaling. Activated CAF promote angiogenesis via secretion of specific pro-angiogenic factors and confer resistance to antiangiogenic therapies [47]. To assess the angiogenic potential of activated fibroblasts, endothelial cell migration was quantified via Transwell assays (Fig. 4E). Quantification showed that TGF-β-activated iMEF-conditioned media (PBS-treated) induced a 1.4-fold increase in CMFDA-labeled sEND.1 cell migration compared to untreated iMEF-CM controls (Fig. 4F). Conditioned media from B7-33-SNPs-treated iMEF cells significantly suppressed sEND.1 migration versus PBS-treated controls. Notably, B7-33-SNPs reduced migration by 57.7 ± 0.9%, demonstrating significantly greater efficacy than free B7-33 (30.0 ± 0.8%), cRGD-B7-33 (27.3 ± 0.6%), or bare SNPs (27.8 ± 10.6%) (Fig. 4G). To further ascertain that the B7-33-SNPs inhibited activated fibroblast-mediated angiogenesis, sprouting angiogenesis assays were conducted. B7-33-SNPs treatment induced a 83.1 ± 12.9% reduction in sprout length compared to PBS-treated controls (Fig. 4H-I). Subsequently, qRT-PCR was employed to analyze the expression levels of activated fibroblast marker proteins and pro-angiogenic factors in iMEFs after B7-33-SNPs treatment. As illustrated in Figure S6, B7-33-SNPs treatment significantly downregulated the expression of Acta2 (α-SMA)‌, Fn1 (fibronectin), Col1a1 and Col1a2 (type I collagen subunits), key markers of activated fibroblasts, along with ‌angiogenic growth factors‌ such as Cxcl12 (CXCL12), Vegfa (VEGFA), Tgfb1 (TGF-β)‌ and Hgf (HGF) (Figure S6). These findings demonstrate that B7-33-SNPs not only reverse iMEF activation but also suppress activated iMEF-driven angiogenesis, thereby enhancing antiangiogenic therapeutic efficacy. Fig. 4Effect of B7-33-SNPs on CAF activation and CAF-mediated angiogenesis. (A) Representative images of iMEF cell or TGF-β-activated iMEF contraction after 48 h of treatment with different formulations. (B) Quantitative analysis of iMEF cell contraction ratios, normalized to the TGF-β control group. (C) Western blotting assessed ECM deposition in iMEF cells or TGF-β-activated iMEF after 48 h of treatment with different formulations. (D) Quantitative analysis of ECM protein levels, normalized to the TGF-β control group. (E) Schematic illustration of the transwell invasion assay setup. sEND.1 cells were seeded into Matrigel-coated transwell inserts and allowed to invade towards iMEF cells or TGF-β-activated iMEF cells treated with different formulations in the lower chamber for 12 h (Created by BioRender.com). (F) CMFDA-stained images of migrated sEND.1 cells in each experimental group. Scale bar: 100 μm. (G) Normalized quantification of migrated cells from CMFDA-stained images, relative to TGF-β control group. (H) Bright-field images of 3D sEND.1 spheroid sprouting on Matrigel following 48 h of co-incubation with corresponding iMEF cells or TGF-β-activated iMEF treated with different formulations. Scale bar:200 μm. Dotted lines denote spheroid sprout boundaries. The magnified view of the white square-marked region in the upper figure is shown below. Scale bar:100 μm. (I) Quantitative analysis of the sprout length from the 3D sEND.1 spheroids, normalized to the TGF-β control group. Data are shown as mean ± SD (n = 6); t-test indicated ‌**p < 0.01 and ***‌p < 0.001 Effect of B7-33-SNPs on CAF activation and CAF-mediated angiogenesis. (A) Representative images of iMEF cell or TGF-β-activated iMEF contraction after 48 h of treatment with different formulations. (B) Quantitative analysis of iMEF cell contraction ratios, normalized to the TGF-β control group. (C) Western blotting assessed ECM deposition in iMEF cells or TGF-β-activated iMEF after 48 h of treatment with different formulations. (D) Quantitative analysis of ECM protein levels, normalized to the TGF-β control group. (E) Schematic illustration of the transwell invasion assay setup. sEND.1 cells were seeded into Matrigel-coated transwell inserts and allowed to invade towards iMEF cells or TGF-β-activated iMEF cells treated with different formulations in the lower chamber for 12 h (Created by BioRender.com). (F) CMFDA-stained images of migrated sEND.1 cells in each experimental group. Scale bar: 100 μm. (G) Normalized quantification of migrated cells from CMFDA-stained images, relative to TGF-β control group. (H) Bright-field images of 3D sEND.1 spheroid sprouting on Matrigel following 48 h of co-incubation with corresponding iMEF cells or TGF-β-activated iMEF treated with different formulations. Scale bar:200 μm. Dotted lines denote spheroid sprout boundaries. The magnified view of the white square-marked region in the upper figure is shown below. Scale bar:100 μm. (I) Quantitative analysis of the sprout length from the 3D sEND.1 spheroids, normalized to the TGF-β control group. Data are shown as mean ± SD (n = 6); t-test indicated ‌**p < 0.01 and ***‌p < 0.001 Tumor-specific accumulation of B7-33-SNPs in vivo Recent evidence underscores the tumor microenvironment as a critical determinant of resistance to antiangiogenic therapy [48, 49]. To characterize the stromal-angiogenic crosstalk in CCA, we performed immunohistofluorescence staining of α-SMA (a marker of CAF) and CD31 (an endothelial cell marker) in paired human CCA and adjacent non-tumor liver tissues from treatment-naive patients. As shown in Fig. 5A-B, both α-SMA-positive stromal cells and CD31-positive microvessels were significantly more abundant in CCA tissues than in normal hepatic parenchyma. Immunohistochemical analysis notably revealed upregulation of relaxin family peptide receptor 1 (RXFP1) in CCA tissues, which colocalized with α-SMA-positive CAF and CD31-positive endothelial cells (Fig. 5A). Consistent with human data, mouse CCA models (HUCCT1 subcutaneous xenografts) demonstrated strong expression of α-SMA, CD31, and RXFP1 in the tumor stroma. Pearson correlation analysis of the TCGA dataset revealed significant positive correlations between gene ACTA2 (α-SMA) and gene PECAM1 (CD31) expression (r = 0.52, p = 0.0013), as well as between gene ACTA2 and gene RXFP1 expression (r = 0.52, p = 0.0012; Fig. 5C). These findings demonstrate a close spatial and functional relationship between CAF, angiogenesis, and RXFP1 signaling in CCA, suggesting that simultaneous disruption of CAF-mediated fibrosis and angiogenesis via targeting RXFP1 may be an effective therapeutic strategy for treating these tumors. For effective cancer therapy, nanoparticles require enhanced circulatory stability and targeted accumulation at disease sites [50, 51]. Previous studies confirmed that B7-33-SNPs extended the circulation time of peptide cRGD-B7-33, improving its in vivo stability [40]. To assess tumor-targeting efficacy, Cy5.5-labeled nanoparticles were subjected to in vivo biodistribution analysis. BALB/c nude mice bearing subcutaneous HUCCT1 xenografts (1 × 106 cells/mouse) received intravenous injections of Cy5.5 dye and Cy5.5-labelled B7-33-SNPs formulations. Ex vivo imaging at 24 h post-injection demonstrated significantly higher tumor accumulation of B7-33-SNPs-Cy5.5 versus control formulations (Fig. 5D). Quantification revealed 2.0-, 2.1-, and 2.1-fold higher tumor fluorescence intensity in B7-33-SNPs-treated mice (0.83 ± 0.07) compared to C7-33-SNPs (0.42 ± 0.04), SNPs (0.38 ± 0.06), and free Cy5.5 (0.39 ± 0.08), respectively (Fig. 5E). Collectively, B7-33-SNPs-Cy5.5 exhibited significantly elevated tumor fluorescence intensity versus controls, which can be ascribed to relaxin R1-mediated targeting and vascular endothelial CD31 binding affinity. These data validate B7-33-SNPs as a dual-functional nanosystem meeting the targeting criteria for CCA therapy. To further investigate nanosystem biodistribution, tumors harvested 24 h post-injection of Cy5.5-labeled formulations were sectioned and analyzed via confocal microscopy. Consistent with ex vivo data, B7-33-SNPs-Cy5.5 exhibited superior tumor accumulation compared to other control groups. B7-33-SNPs distribution correlated with imaging data, showing prominent accumulation at tumor sites (Fig. 5F). Notably, Cy5.5-labeled nanoparticles (red) colocalized with CD31+ endothelial cells (white) and α-SMA+ CAF (green), indicating dual targeting mechanisms via prolonged circulation and Relaxin R1/CD31 interactions. Fig. 5Biodistribution of B7-33-SNPs in a CCA model. (A) Expression and spatial distribution patterns of α-SMA, CD31, and RXFP1 in human CCA tissues and murine CCA cell-derived xenograft (CDX) Models. Representative dual immunofluorescence staining for CD31 (white), α-SMA (red), and RXFP1 (green) in human CCA tissues, adjacent paracancerous tissues, and HUCCT1 subcutaneous xenografts (n = 6). Nuclei were visualized by DAPI counterstaining (blue). Scale bars: 100 μm. (B) Quantitative assessment of normalized protein expression levels for α-SMA, CD31, and RXFP1 in paracancerous tissues and tumor tissues (n = 6) using ImageJ software. (C) Pearson correlation analysis of the Transcripts Per Million (TPM) of gene ACTA2 (α-SMA), gene PECAM1 (CD31), and gene RXFP1 (RXFP1) in human CCA tissues. Correlation coefficients and significance levels are reported. (D) Ex vivo Cy5.5-fluorescence images of tumor tissues 24 h post-injection with B7-33-SNPs, C7-33-SNPs, or SNPs. (E) Quantification of tumor fluorescence intensity. (F) Immunofluorescence localization (CD31 (white)/α-SMA (green)/DAPI (blue)) of Cy5.5-labeled SNPs in tumors (scale bar: 100 μm). Data are shown as mean ± SD (n = 6); t-test indicated ‌**p < 0.01 and ***‌p < 0.001 Biodistribution of B7-33-SNPs in a CCA model. (A) Expression and spatial distribution patterns of α-SMA, CD31, and RXFP1 in human CCA tissues and murine CCA cell-derived xenograft (CDX) Models. Representative dual immunofluorescence staining for CD31 (white), α-SMA (red), and RXFP1 (green) in human CCA tissues, adjacent paracancerous tissues, and HUCCT1 subcutaneous xenografts (n = 6). Nuclei were visualized by DAPI counterstaining (blue). Scale bars: 100 μm. (B) Quantitative assessment of normalized protein expression levels for α-SMA, CD31, and RXFP1 in paracancerous tissues and tumor tissues (n = 6) using ImageJ software. (C) Pearson correlation analysis of the Transcripts Per Million (TPM) of gene ACTA2 (α-SMA), gene PECAM1 (CD31), and gene RXFP1 (RXFP1) in human CCA tissues. Correlation coefficients and significance levels are reported. (D) Ex vivo Cy5.5-fluorescence images of tumor tissues 24 h post-injection with B7-33-SNPs, C7-33-SNPs, or SNPs. (E) Quantification of tumor fluorescence intensity. (F) Immunofluorescence localization (CD31 (white)/α-SMA (green)/DAPI (blue)) of Cy5.5-labeled SNPs in tumors (scale bar: 100 μm). Data are shown as mean ± SD (n = 6); t-test indicated ‌**p < 0.01 and ***‌p < 0.001 Dual-targeting nanotherapy synergistically suppresses CCA growth For many advanced solid tumors, anti-fibrotic and antiangiogenic (e.g., nintedanib, bevacizumab) agents have achieved encouraging results. Although the abundant ECM and increased angiogenesis have been considered attractive therapeutic targets in most clinical studies, single inhibition of them is insufficient to completely suppress CCA progression and prevent the emergence of drug resistance [11–13]. These insights provide a basis for designing more effective therapeutic regimens, wherein the modifiability and functional integration of customized nanomaterials hold promising prospects for simultaneous multi-site targeting. However, while most current nanomedicines focus on regulating CCA stromal fibrosis or angiogenesis individually, the functional integration advantages of nanodrugs for concurrent modulation of tumor stroma and angiogenesis remain largely unexplored [52–54]. Collectively, above findings indicate that targeting CAF activation may delay the generation of antiangiogenic therapy resistance, enabling dual-functional nanovesicles B7-33-SNPs to synergistically inhibit CAF activation and angiogenesis for enhanced antitumor efficacy in the subcutaneous xenograft CCA models. BALB/c nude mice bearing subcutaneous HUCCT1 xenografts were stratified into six groups: PBS, cRGD-B7-33, cRGD-B7-33 + RNPs, SNPs, C7-33-SNPs, and B7-33-SNPs (Fig. 6A). Body weight monitoring revealed comparable tolerability across groups, with no significant inter-group variations (Figure S7). Tumor growth curves demonstrated delayed progression in cRGD-B7-33 and cRGD-B7-33 + RNPs groups versus PBS controls. Although SNPs and C7-33-SNPs further suppressed tumor growth compared to cRGD-B7-33-based treatments, these differences lacked statistical significance (p >0.05). B7-33-SNPs treatment significantly inhibited tumor growth versus SNPs (p < 0.01) and cRGD-B7-33 (p < 0.01) monotherapies, highlighting synergistic efficacy through dual CAF activation and angiogenesis inhibition (Fig. 6B-D). H&E staining of excised tumors showed comparable tissue architecture and necrosis between B7-33-SNPs and PBS groups (Fig. 6E). Western blot analysis confirmed marked reduction of key ECM proteins including collagen I, α-SMA, and fibronectin, following B7-33-SNPs treatment (Fig. 6F-G). Mechanistically, B7-33-SNPs suppressed CXCL12 (SDF-1), a multifunctional mediator secreted by activated CAF in desmoplastic tumors that promotes angiogenesis via vascular endothelial modulation [41]. Quantification revealed 54.3 ± 12.3% CXCL12 downregulation in B7-33-SNPs-treated tumors versus PBS controls. These results collectively validate B7-33-SNPs as a dual-targeting therapeutic that concurrently inhibits CAF activation and angiogenesis to suppress CCA progression. Additionally, while dual-functional nanovesicles B7-33-SNPs concurrently inhibits CAF activation and angiogenesis to suppress subcutaneous xenograft CCA progression, survival curves, complementary experimental models, clinical feasibility, and the durability of long-term effects are also critical issues that future research needs to address. Fig. 6In vivo therapeutic efficacy of B7-33-SNPs against CCA. (A) Schematic depiction of HUCCT1 subcutaneous inoculation and treatment schedule. (B) Growth curves of tumors during the treatment period. (C) Presentation of individual tumor volumes in HUCCT1 CCA models during the therapeutic study. (D) Representative images of tumor-bearing mice at the endpoint, illustrating reduced tumor burden in the B7-33-SNPs group. (E) Tumor sections stained with H&E from mice with different treatments. Scale bar: 200 μm. (F) Western blot analyses of ECM deposition and CXCL12 expression in tumor tissues in indicated groups. (G) Quantification of protein expression, normalized to β-actin, using ImageJ software. Results shown as mean ± SD (n = 6); t-test indicated ‌*p < 0.05, **p < 0.01 and ***‌p < 0.001 In vivo therapeutic efficacy of B7-33-SNPs against CCA. (A) Schematic depiction of HUCCT1 subcutaneous inoculation and treatment schedule. (B) Growth curves of tumors during the treatment period. (C) Presentation of individual tumor volumes in HUCCT1 CCA models during the therapeutic study. (D) Representative images of tumor-bearing mice at the endpoint, illustrating reduced tumor burden in the B7-33-SNPs group. (E) Tumor sections stained with H&E from mice with different treatments. Scale bar: 200 μm. (F) Western blot analyses of ECM deposition and CXCL12 expression in tumor tissues in indicated groups. (G) Quantification of protein expression, normalized to β-actin, using ImageJ software. Results shown as mean ± SD (n = 6); t-test indicated ‌*p < 0.05, **p < 0.01 and ***‌p < 0.001 Immunohistochemical (IHC) analysis was conducted on tumor tissues from B7-33-SNPs-treated and control groups to validate the dual inhibitory effects of B7-33-SNPs on CAF activation and tumor angiogenesis in vivo. IHC analysis demonstrated significant reductions of collagen I, α-SMA, fibronectin, and endothelial marker CD31 in the B7-33-SNPs group compared with the control (Fig. 7A-B). In contrast, cRGD-B7-33-based treatments (cRGD-B7-33 and cRGD-B7-33 + RNPs) elicited modest reductions in these proteins and lacked significant antiangiogenic activity. Monotherapy with SNPs or C7-33-SNPs (bearing VEGFR2/Tie2 decoy receptors) inhibited angiogenesis through sequestration of proangiogenic factors (e.g., VEGFA, ANG2). Collectively, these results establish that B7-33-SNPs uniquely reverse CAF activation and suppress angiogenesis within the tumor stroma, surpassing the efficacy of cRGD-B7-33-based or SNP-containing formulations (e.g., C7-33-SNPs, cRGD-B7-33 + SNPs). Our prior studies demonstrated that ECM overexpression impedes drug delivery [55], and upregulated pro-fibrotic proteins in CAF from antiangiogenic therapy-resistant CCA likely contribute to therapeutic resistance. Attenuating ECM expression may enhance intratumoral perfusion of antiangiogenic agents. To test this hypothesis, Cy5.5-labeled SNPs were utilized to assess nanoparticle perfusion in tumors after 2-week cRGD-B7-33 pretreatment. In vivo fluorescence imaging 1 h post-injection showed enhanced tumor signal clustering in cRGD-B7-33-pretreated mice versus PBS controls (Figures S8A-C). Confocal microscopy corroborated these findings, revealing expansive and intensified Cy5.5 fluorescence in cRGD-B7-33-pretreated tumors. Quantitative analysis confirmed a 2.7-fold increase in nanoparticle penetration depth (9.08 ± 2.23 μm vs. 3.39 ± 1.32 μm; p < 0.001) in cRGD-B7-33-pretreated tumors (Figures S8D-E). Collectively, these data validate that cRGD-B7-33-driven ECM modulation enhances nanoparticle perfusion in CCA xenografts. Fig. 7Immunofluorescence evaluation of stroma modulation in HUCCT1 subcutaneous xenografts. (A) Representative immunofluorescence images showing tumor stromal components: collagen fibers, collagen I, α-SMA, fibronectin, PDGFR-β and CD31. Scale bar: 50 μm. (B) Quantitative analysis of collagen, collagen I, α-SMA, fibronectin, PDGFR-β and CD31 in tumor tissues with different treatments. Results are presented as mean ± SD (n = 6); t-test indicated ‌*p < 0.05, **p < 0.01 and ***‌p < 0.001 Immunofluorescence evaluation of stroma modulation in HUCCT1 subcutaneous xenografts. (A) Representative immunofluorescence images showing tumor stromal components: collagen fibers, collagen I, α-SMA, fibronectin, PDGFR-β and CD31. Scale bar: 50 μm. (B) Quantitative analysis of collagen, collagen I, α-SMA, fibronectin, PDGFR-β and CD31 in tumor tissues with different treatments. Results are presented as mean ± SD (n = 6); t-test indicated ‌*p < 0.05, **p < 0.01 and ***‌p < 0.001 To evaluate the biosafety of B7-33-SNPs, formulations were administered via intravenous injection to BALB/c nude mice. Major organs (heart, liver, spleen, lungs, kidneys) from B7-33-SNPs-treated healthy mice were harvested, sectioned, and subjected to H&E staining. B7-33-SNPs-treated mice exhibited no significant histopathological alterations in paraffin-embedded organ sections versus controls, with no evidence of inflammatory infiltrates or structural abnormalities (Figure S9), demonstrating the biocompatibility of B7-33-SNPs in vital organs. Conclusion The tumor microenvironment (TME), a critical regulator of CCA progression, is dynamically shaped by activated CAF and pathological angiogenesis. CAF within therapy-resistant tumors adopt a pro-fibrotic phenotype characterized by excessive secretion of proangiogenic mediators such as CXCL12 and VEGFA, coupled with aberrant deposition of ECM components including collagen I and fibronectin. These adaptations not only reinforce stromal stiffness but also sustain angiogenic signaling, thereby perpetuating therapeutic resistance. Targeting CAF activation has thus emerged as a strategic imperative to disrupt this self-reinforcing cycle and restore treatment sensitivity. In this study, we developed a biomimetic nanotherapeutic platform (B7-33-SNPs) designed for dual modulation of stromal and vascular compartments of CCA. The system integrates a relaxin receptor-targeting peptide to suppress CAF activation through TGF-β signaling inhibition, alongside endothelial membrane-derived decoy receptors that neutralize key angiogenic ligands. In vivo evaluations demonstrated potent suppression of both CAF-driven desmoplasia and tumor vascularization. Notably, B7-33-SNPs achieved marked tumor growth inhibition without inducing detectable systemic toxicity, underscoring their favorable safety profile. While current clinical paradigms for CCA prioritize surgical resection and cytotoxic chemotherapy, this dual-targeting strategy offers a mechanistically distinct adjuvant approach. By concurrently mitigating stromal barrier formation and angiogenic signaling, it addresses two interdependent hallmarks of therapy resistance. Future investigations should explore synergies with existing modalities and evaluate long-term biodistribution to advance translational applications. Supplementary Information Supplementary Material 1 Supplementary Material 1