Exosomal integrin αvβ3 derived from colorectal cancer promotes the liver pre-metastatic niche formation by inducing cancer-associated fibroblasts activation Authors: Jing Yu, Jiangning Xiang, Lin Yao, Shan Wang, Yifan Liu, Enli Chen, Xiaolin Liu, JinXiang Wu, Yuhao Wang, Lei Zhao Published: 2026 ABSTRACT The formation of the liver pre-metastatic niche (PMN) is a critical factor in the development of colorectal cancer liver metastasis (CRLM), so understanding the mechanism of the liver PMN is crucial for the prevention and treatment of CRLM. Hepatic stellate cells (HSCs) can promote CRLM by forming cancer-associated fibroblasts (CAFs), but the mechanism remains unclear. In this study, we found that integrin αvβ3 is associated with CRLM and the poor clinical prognosis. Exosomal integrin αvβ3 derived from CRC can target HSCs via fibronectin to mediate their activation. We further explored the mechanism and found that exosomal integrin αvβ3 activates the MEK-ERK signaling pathway through the vitronectin-Integrin αvβ3-FAK axis, thereby promoting HSCs activation. After HSCs become CAFs, they promote the formation of the PMN through EMT, angiogenesis, and inflammatory factors, thereby promoting CRLM. In summary, this study clarifies that exosomal integrin αvβ3 derived from CRC can promote the formation of liver PMN by mediating HSC activation into CAFs, providing a novel therapeutic strategy for the prevention and treatment of CRLM.Supplementary InformationThe online version contains supplementary material available at 10.1007/s10238-026-02193-5. Introduction Colorectal cancer (CRC) is the third most common malignant tumor worldwide [1], and more than 50% of colorectal cancer patients will develop metastasis [2, 3]. The 5-year survival rate of colorectal cancer patients who develop distant metastasis is less than 15% [4]. Liver metastasis is particularly prevalent, with a 20–25% incidence rate at initial diagnosis rates [5] and an additional 15–25% developing liver metastasis post-resection [6]. The median survival of untreated CRC patients with liver metastasis is only 6.9 months [6, 7]. The 5-year survival rate of patients with CRLM that cannot be surgically resected is less than 5% [8, 9], while the 5-year survival rate of CRC patients who have complete resection of liver metastasis lesions is 30–57% [8, 10]. Although surgical resection of CRLM can lead to long-term survival, the postoperative recurrence rate is more than 70% [11]. Clinically, 70–80% of patients with CRLM are not suitable for surgical resection [12, 13], and the main treatment options are chemotherapy or chemotherapy combined with targeted therapy [14], though with suboptimal outcomes. Given the suboptimal efficacy of current treatments and the unclear specific mechanisms underlying CRLM, there is an urgent need to explore the potential pathogenic mechanisms, identify novel therapeutic targets, and develop new treatment strategies to improve the prognosis of patients with CRLM. Tumor metastasis is governed by complex interactions between disseminated tumor cells and the host microenvironment, but also affect distant organs to create a microenvironment known as the PMN [15, 16]. Therefore, liver PMN plays an important role in CRLM. The local host stromal microenvironment is one of the key factors involved in the formation of PMN. As an important component of the local stromal microenvironment, CAFs play a key role in PMN formation [17]. CAFs can interact with other cells in the tumor microenvironment through remodeling extracellular matrix (ECM), secreting chemokines and cytokines, and suppressing immunity, thus playing an important role in tumor progression [18, 19]. It has been found that hepatocellular carcinoma cell-derived exosomal miR-1247-3p targets fibroblasts via B4GALT3, activating the integrin β1-NF-κB signaling pathway and converting quiescent fibroblasts into CAFs, thereby promoting lung metastasis [20]. Given the critical role of CAFs in tumor progression, CAFs are considered potential targets for anticancer therapy. HSCs are one of the hepatic stromal cells that can be activated into CAFs in the tumor microenvironment [21, 22]. The activation of HSCs is regulated by multiple signaling pathways, among which the transforming growth factor-β (TGF-β)/Smad pathway serves as a canonical pro-activation cascade. Inflammatory factors secreted by tumor cells maintain the activated phenotype of HSCs via activating the NF-κB and JAK/STAT pathways [23]. The mitogen-activated protein kinase/extracellular signal-regulated kinase (MAPK/ERK) signaling pathway plays a crucial role in HSCs activation. It can be triggered by a variety of upstream signals, including growth factors (such as PDGF and EGF), cytokine, and integrin-mediated focal adhesion signals. Activated ERK drives HSCs activation through phosphorylation of downstream effectors, and promotes the transition of HSCs toward a pro-metastatic phenotype [24–29]. These activated HSCs not only provide physical support for metastatic foci by remodeling ECM, but also secrete a variety of cytokines, chemokines, and pro-angiogenic factors, thereby creating an immunosuppressive, vascular-rich microenvironment that is favorable for the colonization and growth of tumor cells [30]. Therefore, in-depth exploration of the signaling pathways that regulate HSCs activation is of great significance for understanding the molecular mechanisms of liver metastasis and developing therapeutic strategies targeting the PMN. Many studies have shown that HSCs activation is related to CRLM [31, 32], but the specific mechanisms of HSCs activation and PMN formation remain unclear. Exosomes are extracellular vesicles secreted by most cells with a diameter ranging from 30 to 150 nm. The outer layer is a lipid bilayer membrane and carries a variety of substances. These contents can be transferred into the recipient cells to affect the phenotype and function of recipient cells [33–37]. Therefore, these components play an important role in tumor migration and invasion [18]. Integrins are transmembrane cell-surface adhesion receptors present on exosomal membrane. The integrin family contains 18 α subunits and 8 β subunits, which assemble via disulfide bonds to form heterodimers, thus interacting with ECM to activate signaling pathways. The different expression patterns of exosomal integrins represent different organotropism and can be used as biomarkers of organ metastasis [38–41]. For example, exosomal integrin α6β4 and α6β1 promote lung metastasis when taken up by lung fibroblasts and epithelial cells. Exosomal integrin αvβ5 can promote liver metastasis when taken up by kupffer cells [38]. Our preliminary study found high expression of integrin αvβ3 in serum exosomes from patients with CRLM. Integrin αvβ3 has been detected in tumors including breast cancer, liver cancer, prostate cancer and pancreatic cancer [42–45]. Studies have confirmed that integrin αvβ3 plays an important role in tumor progression, drug resistance, angiogenesis, EMT, etc [46]. Integrin αvβ3 can induce EMT and cisplatin resistance in breast cancer [47], and targeting integrin αvβ3 can effectively inhibit the progression of prostate cancer [48]. In addition, Integrin αvβ3 is overexpressed in highly metastatic hepatocellular cancer cell, which can mediate invasion and migration of hepatocellular carcinoma cell [49]. However, the role and mechanism of exosomal integrin αvβ3 in liver PMN of CRC have not been clarified. Integrin αvβ3, also known as the vitronectin receptor, binds to vitronectin in the ECM when integrin αv combines with different integrin β subunits [50, 51]. Studies have shown that different expression patterns of exosomal integrins can bind to resident cells of specific metastatic organs, resulting in directed metastasis. Studies have shown that the liver, as a target organ for metastasis, has a unique microenvironment that can produce serum component, including fibronectin and vitronectin [52]. The ECM component vitronectin is highly expressed in liver metastatic tissues, but not in primary colorectal cancer tissues [52]. As a core component of the ECM, the high expression of vitronectin can induce cross-linking polymerization with other ECM components such as collagen and fibronectin, thereby remodeling the physical state and biological functions of the local hepatic matrix. From the perspective of changes in matrix physical properties, on the one hand, the increased matrix rigidity can provide stable adhesive anchor points for circulating tumor cells (CTCs), facilitating their retention in liver tissue. This mode of action is consistent with the theory that ECM stiffness regulates cell adhesion and migration behavior during tumor metastasis. On the other hand, changes in matrix porosity and permeability can effectively reduce the histological barriers encountered by tumor cells during invasion, thereby accelerating the infiltration of CTCs into the liver parenchyma [53]. Meanwhile, through its RGD domain, vitronectin can bind to integrin receptors (such as αvβ3 and αvβ5) on the surface of HSCs, liver sinusoidal endothelial cells, and tumor cells [54]. It further promotes the continuous deposition and remodeling of ECM components by activating downstream signaling pathways, ultimately facilitating the formation of the pre-metastatic niche. It has shown that activation of the vitronectin-integrin αvβ3/αvβ5 signaling axis can significantly induce HSCs to secrete ECM components such as type I collagen, which not only exacerbates liver fibrosis but also promotes the formation of the pre-metastatic niche [55]. In addition, the synergistic cross-linking effect between vitronectin and fibronectin can significantly enhance the mechanical strength of the matrix and cell adhesion efficiency, creating a favorable microenvironmental condition for the subsequent colonization and proliferation of tumor cells [56]. Therefore, more studies are needed to clarify whether exosomal integrin αvβ3 can target HSCs via vitronectin and the mechanism of liver PMN formation. In this study, we found that the CRC-derived exosomal integrin αvβ3 can target HSCs via binding to vitronectin, thereby activating the FAK/MEK-ERK signaling pathway to promote HSCs activation into CAFs. Activated CAFs subsequently induce EMT, enhance angiogenesis, and secrete proinflammatory factors, collectively remodeling the liver microenvironment to favor metastasis. The study provides a mechanistic insight into CRLM and suggests that exosomal integrin αvβ3 is a new potential target for the prevention and treatment of CRLM. Materials and methods Cell culture Human CRC cell lines HCT116 and SW480, human colorectal epithelial cell line NCM460, human hepatic stellate cell line LX-2, and human umbilical vein endothelial cell line HUVECs were cultured in DMEM medium supplemented with 10% FBS and 1% penicillin-streptomycin. All cells were cultured in an incubator (5%CO2, 37℃). Clinical specimens Clinical specimens were collected from Beijing Friendship Hospital affiliated to Capital Medical University. Inclusion criteria was histologically confirmed CRC in patients aged ≥ 18 years old. The protocol was approved by the hospital’s ethics committee, allowing the collection and analysis of clinical data in accordance with the principles of the Declaration of Helsinki (Approval No. 2025-P2-089-01). Exosome isolation and identification HCT116, SW480 and NCM460 cells were cultured in medium containing 10% exosome-depleted serum (iCell-0650) for 48 h. The conditioned medium (CM) was collected and exosomes were extracted by ultrafiltration and size-exclusion chromatography. The CM was sequentially centrifuged at 300 g for 10 min and 3000 g for 10 min to remove dead cells and cell debris. Subsequently, the supernatant was concentrated at 4000 g using a 100KD ultrafiltration tube at a volume concentration ratio of 1:30. The concentrated cell supernatant was further purified using an Exosupur exclusion column (ECHO BIOTECH, Cat.No.ES910). Fractions 4, 5, 6, 7, and 8 were collected and then concentrated again by centrifugation at 4000 g using a 100KD ultrafiltration tube. The isolated exosomes were aliquoted and stored at −80℃ until subsequent experiments. The concentration and particle size distribution of exosomes were measured by a Flow NanoAnalyzer. The morphology was observed by transmission electron microscopy (TEM). The BCA protein assay kit determined the concentration of exosomes, and then the expression of exosome-specific markers (TSG101, CD81 and CD9) were detected by western bolt. For all subsequent experiments, exosome extracts from different cell lines were normalized to equal protein amounts prior to use. Exosome uptake assay LX-2 cells were seeded in 6-well plates and cultured in serum-free medium for starvation until the cell confluency reached 50%. Exosomes (10ug) were mixed with diluent buffer containing 4uL PKH67 fluorescent dye, followed by incubation at 37℃ for 5 min and subsequent incubation at 4℃ for 15 min to complete the labeling process. After labeling, the PKH67-labeled exosomes were added to the starved LX-2 cells, and the cells were co-incubated for 12 h in an incubator (5% CO2, 37℃). After the co-incubation period, the cells were fixed with 4% paraformaldehyde for 15 min. Subsequently, the cells were stained with DAPI and phalloidin. Finally, the uptake of PKH67-labeled exosomes by LX-2 cells was observed and imaged using a confocal laser scanning microscope. Protein extraction and Western blot analysis Proteins were extracted from cells with RIPA buffer containing PMSF and phosphatase inhibitor. The cell lysates were denatured by adding loading buffer at 100℃ for 5 min. The cell lysates were electrophoresed in SDS-PAGE gel and then transferred to PVDF membranes. The membranes were blocked with 5% non-fat milk in TBST buffer for1h at room temperature prior to incubation with primary specific antibodies overnight at 4℃. After three washes with TBST buffer, the membranes were incubated with HRP-conjugated secondary antibodies for 1 h at 37℃. The protein bands were identified using an ECL system. Additional files: Table S1 contains information on relevant antibodies. RNA extraction and quantitative real-time PCR Total cellular RNA was extracted using the TRIzol reagent, and then total RNA was reverse transcribed into cDNA by using a cDNA Synthesis Kit. Afterwards, quantitative real-time polymerase chain reaction (qRT-PCR) was performed using qPCR SYBR Green Premix reagent. The expression levels of target genes were normalized to GAPDH expression levels, the all reactions were performed in triplicate. The 2−ΔΔCt method was used to calculate the relative gene expression levels. Additional file: Table S2 contains a list of primers. Transwell assay LX-2 cells were suspended in 650ul medium containing 10%FBS and seeded into the lower chamber of 24-well transwell cell culture chambers (8 μm pore size). For cell migration experiment, 2 × 105 HCT116 cells were suspended in 200ul FBS-free medium and seeded into the upper chamber after LX-2 cells were attached to the wall. For cell invasion experiment, 4 × 105 HCT116 cells were suspended in 200ul FBS-free medium and seeded into the matrigel-precoated upper chamber. These cells were incubated with exosomal integrin αvβ3 (20ug/ml), cilengitide (6nM) and exosomal integrin αvβ3 + cilengitide for 36 h. Then, these cells were fixed with 4% paraformaldehyde and stained with 0.1% crystal violet. Finally, these results were photographed under the microscope. ImageJ software was used to count and analyze the stained cells, quantifying the number of migrated and invaded cells. CCK8 and Colony formation assay 5 × 103 HCT116 cells in the logarithmic growth stage were inoculated into 96-well plates. Then, these cells were incubated with exosomal integrin αvβ3, cilengitide and exosomal integrin αvβ3 + cilengitide for 24 h. Then, 10ul CCK8 reagent was added to each well, and the absorbance at 450 nm was measured by microplate reader after 2 h. 1 × 103 HCT116 cells were inoculated into 6-well plates for 2 weeks. After fixation with 4% paraformaldehyde, these cells were stained with 0.1% crystal violet for 30 min. Finally, the colony formation was counted. ImageJ software was used to analyze and quantify the number of cells per colony. Tube formation assay First, all things were precooled at 4℃. 4.5 × 104 HUVECs were treated with exosomes and seeded in Matrigel for 3–8 h. Tube-like structures were imaged under the microscope. Sphere formation assay 1 × 103 colorectal cells were seeded in ultra -low attachment 6-well microplate and incubated for 7–10 days on DMEM medium containing 20ul/ml B27, 20ng/ml EGF, and 20ng/ml bFGF. The diameter of sphere was evaluated under the microscope. ELISA Serum and exosomal integrin αvβ3 levels were measured using ELISA kits (Integrin αv: LSBio LS-F7188; Integrin β3: LSBio LS-F21226) following the manufacturer’s protocols. The OD values were recorded at 450 nm, and concentrations were determined based on standard curves. Lentiviral transduction To establish the reliable knockdown and overexpression cell lines, we used the lentiviral vectors to create integrinαvβ3 shRNA and overexpression plasmid. These plasmids were provided by Beijing Likeli Biotechnology Co., Ltd. The constructed plasmid were co-transfected into 293T cells for 48 h with the viral packaging plasmids psPAX2 and pMD2.G. The lentivirus supernatant was collected and filtered by 0.45 μm filter. Then, the lentivirus supernatant was cultured with HCT116 cells and SW480 cells for 24 h. Puromycin and Blasticidin (5ug/ml) selection were applied to the cells. The drug-containing medium was replaced every 2–3 days, and selection was maintained for 7–14 days. The overexpression and knockdown efficiencies were verified using qPCR and WB. Additional file: Table S3-4 show the target sequences for specific genes and the primers for overexpressed genes. Statistical analysis Date analysis used the Graphpad Prism8 and R 4.4.3 software, and the experiments repeated at least 3 times. The data results were represented by mean ± standard, and one-way analysis of variance or t-test analysis were used for analysis. ROC curves were plotted using the pROC package in R software. For the missing data corresponding to “unknown” in Tables 1 and 2, we strictly followed the standard principles for handling missing values in clinical statistical analysis: these data were excluded from the statistical analysis of the corresponding indicators, and no data imputation was performed to avoid bias caused by artificial assumptions. p < 0.05 was statistically significant.Table 1Correlation between integrin αv expression and clinical pathology characteristics in CRCVariableITGαv expressionLow(n = 22)High(n = 17)X2P valueAge≥ 6018(81.8)14(82.4)<0.0011.000<604(18.2)3(17.6)GenderFemale10(45.5)4(23.5)1.1640.281Male12(54.5)13(76.5)TNMI-III18(81.8)5(29.5)8.8270.003**IV4(18.2)12(70.5)BMI<2412(54.5)9(52.9)<0.0011.000≥ 2410(45.5)8(47.1)Microsatellite StatusMSI5(22.7)1(5.9)0.3570.550MSS15(68.2)10(58.8)Unknown2(9.1)6(35.3)Lymphatic vessel invasionYes6(27.3)4(23.5)0.1370.712No12(54.5)4(23.5)Unknown4(18.2)9(53.0)Blood vessel invasionYes7(31.8)4(23.5)0.0100.921No11(50.0)4(23.5)Unknown4(18.2)9(53.0)Perineural invasionYes5(22.7)5(29.4)0.0100.921No13(59.1)3(17.6)Unknown4(18.2)9(53.0)Her2+8(36.4)4(23.5)0.0100.921-11(50.0)6(35.3)Unknown3(13.6)7(41.2)RAS mutation+11(50.0)4(23.5)0.5740.449-8(36.4)7(41.2)Unknown3(13.6)6(35.3)BRAF mutation+0(0.0)0(0.0)2.1330.144-19(86.4)11(64.7)Unknown3(13.6)6(23.5)Ki67<75%12(54.5)4(23.5)<0.0011.000≥ 75%6(27.3)3(17.6)Unknown4(18.2)10(58.9)Table 2Correlation between integrin β3 expression and clinical pathology characteristics in CRCVariableITGβ3 expressionLow(n = 15)High(n = 24)X2P valueAge≥ 6013(86.7)19(79.2)0.0270.869<602(13.3)5(20.8)GenderFemale4(26.7)10(41.7)0.3680.544Male11(73.3)14(58.3)TNMI-III13(86.7)10(41.7)5.9780.014*IV2(13.3)14(58.3)BMI<249(60.0)12(50.0)0.0780.780≥ 246(40.0)12(50.0)Microsatellite StatusMSI4(26.7)2(8.3)0.2950.587MSS11(73.3)14(58.3)Unknown0(0.0)8(33.4)Lymphatic vessel invasionYes5(33.3)5(20.8)0.0480.826No10(66.7)6(25.0)Unknown0(0.0)13(54.2)Blood vessel invasionYes5(33.3)6(25.0)0.4620.497No10(66.7)5(20.8)Unknown0(0.0)13(54.2)Perineural invasionYes3(80.0)7(29.1)0.4620.497No12(20.0)4(16.7)Unknown0(0.0)13(54.2)Her2+7(46.7)5(20.8)0.4620.497-7(46.7)10(41.7)Unknown1(6.6)9(37.5)RAS mutation+7(46.7)8(33.3)<0.0011.000-7(46.7)8(33.3)Unknown1(6.6)8(33.3)BRAF mutation+0(0.0)0(0.0)0.1330.7715-14(93.3)16(66.7)Unknown1(6.7)8(33.3)Ki67<75%9(60.0)7(29.1)<0.0011.000≥ 75%5(33.3)4(16.7)Unknown1(6.7)13(54.2) Table 1Correlation between integrin αv expression and clinical pathology characteristics in CRCVariableITGαv expressionLow(n = 22)High(n = 17)X2P valueAge≥ 6018(81.8)14(82.4)<0.0011.000<604(18.2)3(17.6)GenderFemale10(45.5)4(23.5)1.1640.281Male12(54.5)13(76.5)TNMI-III18(81.8)5(29.5)8.8270.003**IV4(18.2)12(70.5)BMI<2412(54.5)9(52.9)<0.0011.000≥ 2410(45.5)8(47.1)Microsatellite StatusMSI5(22.7)1(5.9)0.3570.550MSS15(68.2)10(58.8)Unknown2(9.1)6(35.3)Lymphatic vessel invasionYes6(27.3)4(23.5)0.1370.712No12(54.5)4(23.5)Unknown4(18.2)9(53.0)Blood vessel invasionYes7(31.8)4(23.5)0.0100.921No11(50.0)4(23.5)Unknown4(18.2)9(53.0)Perineural invasionYes5(22.7)5(29.4)0.0100.921No13(59.1)3(17.6)Unknown4(18.2)9(53.0)Her2+8(36.4)4(23.5)0.0100.921-11(50.0)6(35.3)Unknown3(13.6)7(41.2)RAS mutation+11(50.0)4(23.5)0.5740.449-8(36.4)7(41.2)Unknown3(13.6)6(35.3)BRAF mutation+0(0.0)0(0.0)2.1330.144-19(86.4)11(64.7)Unknown3(13.6)6(23.5)Ki67<75%12(54.5)4(23.5)<0.0011.000≥ 75%6(27.3)3(17.6)Unknown4(18.2)10(58.9) Correlation between integrin αv expression and clinical pathology characteristics in CRC Table 2Correlation between integrin β3 expression and clinical pathology characteristics in CRCVariableITGβ3 expressionLow(n = 15)High(n = 24)X2P valueAge≥ 6013(86.7)19(79.2)0.0270.869<602(13.3)5(20.8)GenderFemale4(26.7)10(41.7)0.3680.544Male11(73.3)14(58.3)TNMI-III13(86.7)10(41.7)5.9780.014*IV2(13.3)14(58.3)BMI<249(60.0)12(50.0)0.0780.780≥ 246(40.0)12(50.0)Microsatellite StatusMSI4(26.7)2(8.3)0.2950.587MSS11(73.3)14(58.3)Unknown0(0.0)8(33.4)Lymphatic vessel invasionYes5(33.3)5(20.8)0.0480.826No10(66.7)6(25.0)Unknown0(0.0)13(54.2)Blood vessel invasionYes5(33.3)6(25.0)0.4620.497No10(66.7)5(20.8)Unknown0(0.0)13(54.2)Perineural invasionYes3(80.0)7(29.1)0.4620.497No12(20.0)4(16.7)Unknown0(0.0)13(54.2)Her2+7(46.7)5(20.8)0.4620.497-7(46.7)10(41.7)Unknown1(6.6)9(37.5)RAS mutation+7(46.7)8(33.3)<0.0011.000-7(46.7)8(33.3)Unknown1(6.6)8(33.3)BRAF mutation+0(0.0)0(0.0)0.1330.7715-14(93.3)16(66.7)Unknown1(6.7)8(33.3)Ki67<75%9(60.0)7(29.1)<0.0011.000≥ 75%5(33.3)4(16.7)Unknown1(6.7)13(54.2) Correlation between integrin β3 expression and clinical pathology characteristics in CRC Results High expression of integrin αvβ3 is associated with liver metastasis and poor prognosis in CRC In 3 pairs of tissue specimens from CRC patients, the expression of integrin αvβ3 in tumor tissues was significantly higher than paracancerous tissue (Fig. 1A). To explore the correlation between integrin αvβ3 expression and patient prognosis, survival analyses were performed using the Kaplan-Meier (KM) Plotter database (https://kmplot.com/analysis/). Results from the KM Plotter database showed that CRC patients with high integrin αvβ3 expression had significantly shorter overall survival (OS) compared to those with low expression. Consistent with this, survival curve analysis revealed that high integrin αvβ3 expression was associated with significantly reduced disease-free survival (DFS) (Fig. 1B), collectively suggesting that elevated integrin αvβ3 expression correlates with poor prognosis in CRC. To verify the relationships between integrin αvβ3 and CRLM, we collected 39 serum specimens from CRC patients, including patients with liver metastasis (n = 10) and without liver metastasis (n = 29) (Tables 1 and 2). We detected the expression of integrin αvβ3 in serum by ELISA. We found that serum integrin αvβ3 levels were significantly higher in CRC patients with liver metastasis than in those without liver metastasis, regardless of the presence of extrahepatic metastasis (Fig. 1C). In addition, the expression of integrin αvβ3 was positively correlated with tumor stage, suggesting that patients with high expression of integrin αvβ3 had poor prognosis (Fig. 1D). Then, we calculated the integrin αvβ3 cutoff value of patients with or without liver metastasis, and divide them into high or low expression groups according to the cutoff value. We found that the liver metastasis rate in the high expression group is higher than that in the low expression group (Fig. 1E-F). We extracted serum exosomes from CRC patients. We found that integrin αvβ3 was also highly expressed in the serum exosomes of CRC patients with liver metastasis (Fig. 1G). Notably, analysis of serum exosomes showed even higher specificity and sensitivity for detecting liver metastasis, as demonstrated by an increased AUC in ROC curve analysis (Fig. 1H). Therefore, serum exosomal integrin αvβ3 has better sensitivity and specificity, which can improve the accuracy of clinical detection. In summary, high expression of integrin αvβ3 is associated with liver metastasis and poor prognosis of CRC.Fig. 1High expression of integrin αvβ3 is associated with liver metastasis and poor clinical prognosis of CRC. (A) The expression of integrin αv and integrin β3 in tumor tissues and paracancerous tissues of CRC patients were detected by western bolt. (B) The OS and DFS of patients with different expression of integrin αvβ3. (C) Elisa was performed to detect the serum integrin αvβ3 expression in CRC patients with and without liver metastasis (non-meta + other-meta). (D) Integrin αvβ3 expression in the serum of CRC patients with different tumor stages. E-F. The cutoff value was defined as the diagnostic threshold corresponding to the point that maximizes Youden’s Index (Youden’s Index=Sensitivity+Specificity-1). Comparison of CRLM rates with high and low expression of integrin αvβ3. G. Elisa was used to assess the expression of serum exosomal integrin αv and integrin β3 in CRC patients with and without liver metastasis. H. ROC curves were used to compare the accuracy of serum integrin αvβ3 and serum exosomal integrin αvβ3 in the diagnosis of CRLM High expression of integrin αvβ3 is associated with liver metastasis and poor clinical prognosis of CRC. (A) The expression of integrin αv and integrin β3 in tumor tissues and paracancerous tissues of CRC patients were detected by western bolt. (B) The OS and DFS of patients with different expression of integrin αvβ3. (C) Elisa was performed to detect the serum integrin αvβ3 expression in CRC patients with and without liver metastasis (non-meta + other-meta). (D) Integrin αvβ3 expression in the serum of CRC patients with different tumor stages. E-F. The cutoff value was defined as the diagnostic threshold corresponding to the point that maximizes Youden’s Index (Youden’s Index=Sensitivity+Specificity-1). Comparison of CRLM rates with high and low expression of integrin αvβ3. G. Elisa was used to assess the expression of serum exosomal integrin αv and integrin β3 in CRC patients with and without liver metastasis. H. ROC curves were used to compare the accuracy of serum integrin αvβ3 and serum exosomal integrin αvβ3 in the diagnosis of CRLM Exosomal integrin αvβ3 is highly expressed in CRC cells with high liver metastatic potential In order to investigate the mechanism of CRLM, we selected three cell lines, including human CRC cell HCT116 with high liver metastatic potential, human CRC cell SW480 with low liver metastatic potential, and human normal colorectal epithelial cell NCM460 [57]. Firstly, we extracted exosomes from these cell supernatants by ultrafiltration and size-exclusion chromatography. Exosomes showed a “cup-like” or “disc-like” structure under transmission electron microscopy (TEM) (Fig. 2A). Flow NanoAnalyzer analysis revealed that exosomes had a particle size range of 30–150 nm (Fig. 2B). Exosomes were validated by western blot using the exosome-specific markers, such as TSG101, CD81, and CD9 (Fig. 2C). In order to explore the relationship between exosomal integrin αvβ3 and CRLM, we detected the expression of exosomal integrin αvβ3 in CRC cells with different liver metastatic potentials by western blot. We found that the expression of integrin αvβ3 in HCT116 cell-derived exosomes was higher than that in SW480 cell-derived exosomes (Fig. 2D). Therefore, high expression of exosomal integrin αvβ3 is associated with enhanced liver metastatic potential of CRC cells.Fig. 2Exosomal integrin αvβ3 is highly expressed in highly liver-metastatic CRC cells. (A) TEM showed the typical structures of exosomes isolated from HCT116, SW480, NCM460 cells. Scale bar=200 nm. (B) The particle size of exosomes was detected by FCM (30–150 nm). (C) The exosome-specific markers were detected by western blot. (D) The expression of integrin αvβ3 in exosomes derived from NCM460, HCT116, SW480 cells were detected by western blot. The experiments were repeated at least three times and all data are presented as mean ± standard deviation. (* p < 0.05, ** p < 0.01, *** p < 0.001, **** p < 0.0001) Exosomal integrin αvβ3 is highly expressed in highly liver-metastatic CRC cells. (A) TEM showed the typical structures of exosomes isolated from HCT116, SW480, NCM460 cells. Scale bar=200 nm. (B) The particle size of exosomes was detected by FCM (30–150 nm). (C) The exosome-specific markers were detected by western blot. (D) The expression of integrin αvβ3 in exosomes derived from NCM460, HCT116, SW480 cells were detected by western blot. The experiments were repeated at least three times and all data are presented as mean ± standard deviation. (* p < 0.05, ** p < 0.01, *** p < 0.001, **** p < 0.0001) The exosomal integrin αvβ3 targets HSCs via vitronectin to mediate HSCs activation Exosomal integrins can interact with specific resident cells in target metastatic organ to alter the biological properties of these cells, thus promoting tumor metastasis [38, 41]. Given that high expression of exosomal integrin αvβ3 is associated with CRLM, we investigated whether the exosomal integrin αvβ3 mediates HSCs activation. First, Confocal microscopy demonstrated that PKH67-labeled exosomes from CRC cells were internalized by LX-2 cells (Fig. 3A). It is well established that HSCs can be activated into CAFs in response to changes in the tumor microenvironment [58, 59]. Spearman correlation analysis using TIMER2.0 with data from the TCGA-COAD dataset (n = 458) showed that integrin αvβ3 expression is positively associated with CAFs infiltration (Fig. 3B). In addition, we found that co-incubation of LX-2 cells with exosomal integrin αvβ3 upregulated the expression of CAFs markers (α-SMA, FAP) (Fig. 3C). To further explore how exosomal integrin αvβ3 targets HSCs activation, we performed an analysis using the STRING protein-protein interaction database. It suggested a close interaction between integrin αvβ3 and vitronectin (VTN, marked by the red circle) (Fig. 3D). Integrin αvβ3 functions as a receptor for vitronectin, when αv subunits from heterodimers with β subunits, they can bind to vitronectin in the ECM, thereby promoting cell adhesion and diffusion. We found that co-incubation of LX-2 cells with exosomal integrin αvβ3 increased the expression of vitronectin (Fig. 3E). Cilengitide is an inhibitor that prevents integrins from binding to vitronectin. Since HCT116 cells secrete exosomes with high integrin αvβ3 expression, we used HCT116-derived exosomes for functional validation experiments. As expected, co-incubation of LX-2 cells with HCT116 cell-derived exosomes resulted in increased expression of CAFs markers. We found that the expression of α-SMA and FAP in LX-2 cells decreased after the addition of the inhibitor cilengitide (Fig. 3F). These results suggested that the exosomal integrin αvβ3 activates HSCs into CAFs by targeting vitronectin on HSCs.Fig. 3The exosomal integrin αvβ3 targets to HSCs via vitronectin to mediate HSCs activation. (A) Immunofluorescence assay was used to detect the internalization of PKH67-labeled exosomes by LX-2 cells. Scale bar = 20 μm. (B) Correlation between integrin αvβ3 expression and CAFs infiltration based on TIMER2.0. (C) The expression of CAFs markers (α-SMA, FAP) in LX-2 following 48 h co-incubation with exosomes from SW480 and HCT116 cells was detected by western blot. (D) Proteins associated with integrin αvβ3 were analyzed by STRING. (E) The vitronectin expression in LX-2 cells after 48 h co-incubation with exosomes derived from SW480 and HCT116 cells was detected by western blot. (F) The expression of CAFs markers (α-SMA、FAP) in LX-2 cells following 48 h co-incubation with HCT116-exo, cilengitide, and HCT116-exo+cilengitide respectively. The experiments were repeated at least three times and all data are presented as mean ± standard deviation. (* p < 0.05, ** p < 0.01, *** p < 0.001, **** p < 0.0001) The exosomal integrin αvβ3 targets to HSCs via vitronectin to mediate HSCs activation. (A) Immunofluorescence assay was used to detect the internalization of PKH67-labeled exosomes by LX-2 cells. Scale bar = 20 μm. (B) Correlation between integrin αvβ3 expression and CAFs infiltration based on TIMER2.0. (C) The expression of CAFs markers (α-SMA, FAP) in LX-2 following 48 h co-incubation with exosomes from SW480 and HCT116 cells was detected by western blot. (D) Proteins associated with integrin αvβ3 were analyzed by STRING. (E) The vitronectin expression in LX-2 cells after 48 h co-incubation with exosomes derived from SW480 and HCT116 cells was detected by western blot. (F) The expression of CAFs markers (α-SMA、FAP) in LX-2 cells following 48 h co-incubation with HCT116-exo, cilengitide, and HCT116-exo+cilengitide respectively. The experiments were repeated at least three times and all data are presented as mean ± standard deviation. (* p < 0.05, ** p < 0.01, *** p < 0.001, **** p < 0.0001) Modulation of exosomal integrin αvβ3 alters HSCs activation To further confirm the role of exosomal integrin αvβ3 in HSCs activation, we constructed integrin αvβ3-knockdown HCT116 cells (Fig. 4A-B). We found that the co-incubation of LX-2 cells with exosomes derived from integrin αvβ3-knockdown HCT116 cells resulted in reduced expression of vitronectin, α-SMA and FAP compared to exosomes derived from HCT116 cells (Fig. 4C-D). Additionally, we also constructed integrin αvβ3-overexpressing SW480 cells (Fig. 4E-F). Compared with SW480 cell-derived exosomes, the co-incubation of LX-2 cells with exosomes derived from integrin αvβ3-overexpressing SW480 cells increased the expression of vitronectin, α-SMA and FAP (Fig. 4G-H). In summary, the exosomal integrin αvβ3 can target HSCs and activate HSCs to CAFs via vitronectin.Fig. 4Knockdown of exosomal integrin αvβ3 inhibits HSCs activation, while overexpression of exosomal integrin αvβ3 promotes HSCs activation. A-B. Q-PCR and western-blot were performed to detect the efficiency of shRNA targeting αvβ3 in HCT116 cells. C. Identification of exosomes derived from integrin αvβ3-knockdown HCT116 cells by western-blot, TEM and FCM. D. Western blot was utilized to analyze the vitronectin, α-SMA and FAP expression in LX-2 cells following 48 h co-incubation with HCT116-exo and HCT116 KD-exo. E-F. Q-PCR and western-blot were performed to detect the overexpression efficiency of integrin αvβ3 in SW480 cells. G. Identification of exosomes derived from integrin αvβ3-overexpressiong SW480 cells by western-blot, TEM and FCM. H. Western blot was utilized to analyze the vitronectin, α-SMA and FAP expression in LX-2 cells following 48 h co-incubation with SW480-exo and SW480 OE-exo. The experiments were repeated at least three times and all data are presented as mean ± standard deviation. (* p < 0.05, ** p < 0.01, *** p < 0.001, **** p < 0.0001) Knockdown of exosomal integrin αvβ3 inhibits HSCs activation, while overexpression of exosomal integrin αvβ3 promotes HSCs activation. A-B. Q-PCR and western-blot were performed to detect the efficiency of shRNA targeting αvβ3 in HCT116 cells. C. Identification of exosomes derived from integrin αvβ3-knockdown HCT116 cells by western-blot, TEM and FCM. D. Western blot was utilized to analyze the vitronectin, α-SMA and FAP expression in LX-2 cells following 48 h co-incubation with HCT116-exo and HCT116 KD-exo. E-F. Q-PCR and western-blot were performed to detect the overexpression efficiency of integrin αvβ3 in SW480 cells. G. Identification of exosomes derived from integrin αvβ3-overexpressiong SW480 cells by western-blot, TEM and FCM. H. Western blot was utilized to analyze the vitronectin, α-SMA and FAP expression in LX-2 cells following 48 h co-incubation with SW480-exo and SW480 OE-exo. The experiments were repeated at least three times and all data are presented as mean ± standard deviation. (* p < 0.05, ** p < 0.01, *** p < 0.001, **** p < 0.0001) Exosomal integrin αvβ3 activates the FAK/MEK-ERK pathway in HSCs In order to determine the mechanism by which exosomal integrin αvβ3 activate HSCs, we found that the expression of integrin αvβ3 increased after LX-2 cells co-incubation with exosomes (Fig. 5A). Notably, this upregulation was reversed when cilengitide was added to the co-culture system of LX-2 cells and HCT116 cell-derived exosomes (Fig. 5B). Additionally, co-incubation of LX-2 cells with exosomes from integrin αvβ3-knockdown HCT116 cells led to reduced integrin αvβ3 expression (Fig. 5C). However, after co-incubation with exosomes derived from integrin αvβ3-overexpressing SW480 cells, the expression of integrin αvβ3 in LX-2 cells was up-regulated (Fig. 5D). Therefore, CRC cell-derived exosomes can deliver integrin αvβ3 to HSCs, promoting HSCs activation. Many studies have shown that integrins primarily trigger the FAK signaling pathway, a non-receptor kinase that is responsible for cell adhesion and migration, thus regulating various biological functions. FAK can play a role in cancer progression through cytoplasmic kinase activity, which is largely dependent on integrin signaling [60]. Therefore, we further explore whether the CRC cell-derived exosomal integrin αvβ3 activates the FAK signaling pathway in HSCs. We found that the expression of p-FAK in LX-2 cell significantly increased compared to the control group, and the expression decreased after adding cilengitide (Fig. 5E). However, the signaling pathway of exosomal integrin αvβ3 involved in HSCs activation is still unknown. To further dissect the downstream signaling cascade, we found that the expression of p-ERK and p-MEK in LX-2 cells increased after co-incubation with exosomal integrin αvβ3, while the expression can be reversed after adding cilengitide (Fig. 5E). In summary, our results suggested that exosomal integrin αvβ3 promotes HSCs activation by activating the FAK/MEK-ERK pathway.Fig. 5Exosomal integrin αvβ3 activates the FAK/MEK-ERK pathway in HSCs. (A) Western blot analysis of the integrin αvβ3 expression in LX-2 cells following 48 h co-incubation with SW480-exo and HCT116-exo. (B) The integrin αvβ3 expression in LX-2 cells was analyzed by western blot after 48 h co-incubation with HCT116-exo, cilengitide, and HCT116-exo+cilengitide respectively. C-D. The integrin αvβ3 expression in LX-2 cells was analyzed by western blot after 48 h co-incubation with HCT116-exo, HCT116 KD-exo, SW480-exo, SW480 OE-exo. E. Western blot analysis of the expression of p-FAK/FAK, p-ERK/ERK, and p-MEK/MEK in LX-2 cells after 48 h co-incubation with exosomal ITGαvβ3, cilengitide, and exosomal ITGαvβ3 + cilengitide respectively. The experiments were repeated at least three times and all data are presented as mean ± standard deviation. (* p < 0.05, ** p < 0.01, *** p < 0.001, **** p < 0.0001) Exosomal integrin αvβ3 activates the FAK/MEK-ERK pathway in HSCs. (A) Western blot analysis of the integrin αvβ3 expression in LX-2 cells following 48 h co-incubation with SW480-exo and HCT116-exo. (B) The integrin αvβ3 expression in LX-2 cells was analyzed by western blot after 48 h co-incubation with HCT116-exo, cilengitide, and HCT116-exo+cilengitide respectively. C-D. The integrin αvβ3 expression in LX-2 cells was analyzed by western blot after 48 h co-incubation with HCT116-exo, HCT116 KD-exo, SW480-exo, SW480 OE-exo. E. Western blot analysis of the expression of p-FAK/FAK, p-ERK/ERK, and p-MEK/MEK in LX-2 cells after 48 h co-incubation with exosomal ITGαvβ3, cilengitide, and exosomal ITGαvβ3 + cilengitide respectively. The experiments were repeated at least three times and all data are presented as mean ± standard deviation. (* p < 0.05, ** p < 0.01, *** p < 0.001, **** p < 0.0001) Activated HSCs by exosomal integrin αvβ3 promote the formation of a liver PMN Exosomes can change the biological behavior of tumor cells. In order to explore the role of exosomal integrin αvβ3 in regulating the malignant behaviors of CRC cells, we analyzed cell proliferation, migration, invasion and sphere-forming capacities through CCK8, colony formation, transwell and sphere formation assays. Specifically, LX-2 cells were treated with exosomal integrin αvβ3, cilengitide, or exosomal integrin αvβ3 + cilengitide for 48 h. The conditioned medium was then collected and co-incubated with HCT116 cells. (Fig. 6A). The CCK8 and colony formation assays showed that exosomal integrin αvβ3 enhanced the proliferation capacity of CRC cells compared with the control group, while the inhibitor group significantly decreased cell viability and colony formation efficiency (Fig. 6B-C). Transwell assay showed that the exosomal integrin αvβ3 can promote CRC cell migration and invasion (Fig. 6D). Sphere formation assay showed that the exosomal integrin αvβ3 can enhance the ability of sphere-forming capacity of CRC cells, indicating increased cancer stem cell (CSC)-like properties (Fig. 6E). In general, exosomal integrin αvβ3, after activating HSCs, can regulate the malignant behaviors of CRC cells when co-cultured with them, thereby contributing to CRC progression. The formation of PMN is a complex process. The primary tumor creates a microenvironment conducive to the growth of circulating tumor cells by secreting some factors before distant metastasis [61]. We further explore how to induce the formation of PMN after the formation of CAFs. Some studies have found that CAFs can crosstalk with tumor cells and form a favorable microenvironment for tumor invasion and metastasis by secreting cytokines, EMT, angiogenesis, inflammation and immunosuppression [62, 63]. To characterize the microenvironmental changes induced by CAFs, we analyzed key molecular markers in the co-culture system. We found that the expression of N-cadherin and vimentin in HCT116 cells is up-regulated and the expression of E-cadherin is down-regulated after LX-2 cells co-cultured with exosomal integrin αvβ3 and HCT116 cells (Fig. 6F). This demonstrated that exosomal integrin αvβ3-activated HSCs promote the EMT pathway of CRC cells, providing direct evidence that CAFs can facilitate liver metastasis through the EMT pathway. Furthermore, we detected the secretion of inflammatory cytokines in the supernatant of LX-2 cells following co-incubation with exosomal integrin αvβ3. The expression of interleukin-8 (IL-8), interleukin-6 (IL-6), and transforming growth factor-β (TGF-β) was significantly increased, and this effect was abrogated by cilengitide treatment (Fig. 6G). It suggested that CAFs can promote the pre-metastatic inflammatory microenvironment conducive to metastasis. In order to detect the effect of exosomal integrin αvβ3 on angiogenesis, we conducted tube formation assay. We found that exosomal integrin αvβ3 group significantly promoted angiogenesis, while the inhibitor cilengitide group significantly inhibited angiogenesis (Fig. 6H). In addition, exosomal integrin αvβ3 can up-regulate the expression of pro-angiogenesis related genes VEGF and FLT-1, and the inhibitor cilengitide can reverse the expression of VEGF and FLT-1 (Fig. 6I). These results suggest that exosomal integrin αvβ3 promotes angiogenesis by up-regulating the expression of genes related to angiogenesis. Overall, the activation of HSCs into CAFs can promote CRLM by promoting EMT, promoting the formation of inflammatory microenvironment, and promoting angiogenesis to induce changes in the PMN.Fig. 6Activated HSCs by exosomal integrin αvβ3 promote the formation of a liver PMN. A. Schematic diagram of the experimental design: LX-2 cells were treated with exosomal αvβ3, cilengitide, or exosomal integrin αvβ3 + cilengitide for 48 h. Conditioned medium was then collected and co-incubated with HCT116 cells. B-C. The proliferation capacity of HCT116 cells was analyzed by CCK8 and colony formation assays. D. The migration and invasion abilities of HCT116 cells were analyzed by transwell assays. Scale bar = 100 μm. E. Spheroid-forming ability of HCT116 cells was analyzed by sphere formation assay. Scale bar = 100 μm. F. Western blot was conducted to detect the expression of EMT-related proteins (E-cadherin, N-cadherin and vimentin) in HCT116 cells following 48 h co-incubation with conditioned medium from LX-2 cells or LX-2 cells treated with exosomal integrin αvβ3. G. ELISA was used to measure the levels of inflammatory cytokins (IL-8, IL-6, and TGF-β) in the cell supernatants after 48 h co-incubation with exosomal integrin αvβ3 cilengitide, and exosomal integrin αvβ3 + cilengitide. H. Tube formation assay was performed to evaluate the angiogenic capacity of HUVECs cultured with conditioned medium from LX-2 cells, LX-2 cells treated with exosomal integrin αvβ3, or LX-2 cells co-treated with exosomal integrin αvβ3 and cilengitide. I. Western blot was used to examine the expression of angiogenesis-related proteins (VEGF, FLT-1) in HCT116 cells following 48 h co-incubation with conditioned medium from LX-2 cells, LX2 cells treated with exosomal integrin αvβ3, LX-2 cells treated with cilengitide, or LX-2 cells co-treated with exosomal integrin αvβ3 and cilengitide. The experiments were repeated at least three times and all data are presented as mean ± standard deviation. (* p < 0.05, ** p < 0.01, *** p < 0.001, **** p < 0.0001) Activated HSCs by exosomal integrin αvβ3 promote the formation of a liver PMN. A. Schematic diagram of the experimental design: LX-2 cells were treated with exosomal αvβ3, cilengitide, or exosomal integrin αvβ3 + cilengitide for 48 h. Conditioned medium was then collected and co-incubated with HCT116 cells. B-C. The proliferation capacity of HCT116 cells was analyzed by CCK8 and colony formation assays. D. The migration and invasion abilities of HCT116 cells were analyzed by transwell assays. Scale bar = 100 μm. E. Spheroid-forming ability of HCT116 cells was analyzed by sphere formation assay. Scale bar = 100 μm. F. Western blot was conducted to detect the expression of EMT-related proteins (E-cadherin, N-cadherin and vimentin) in HCT116 cells following 48 h co-incubation with conditioned medium from LX-2 cells or LX-2 cells treated with exosomal integrin αvβ3. G. ELISA was used to measure the levels of inflammatory cytokins (IL-8, IL-6, and TGF-β) in the cell supernatants after 48 h co-incubation with exosomal integrin αvβ3 cilengitide, and exosomal integrin αvβ3 + cilengitide. H. Tube formation assay was performed to evaluate the angiogenic capacity of HUVECs cultured with conditioned medium from LX-2 cells, LX-2 cells treated with exosomal integrin αvβ3, or LX-2 cells co-treated with exosomal integrin αvβ3 and cilengitide. I. Western blot was used to examine the expression of angiogenesis-related proteins (VEGF, FLT-1) in HCT116 cells following 48 h co-incubation with conditioned medium from LX-2 cells, LX2 cells treated with exosomal integrin αvβ3, LX-2 cells treated with cilengitide, or LX-2 cells co-treated with exosomal integrin αvβ3 and cilengitide. The experiments were repeated at least three times and all data are presented as mean ± standard deviation. (* p < 0.05, ** p < 0.01, *** p < 0.001, **** p < 0.0001) Discussion Studies have confirmed that integrins play an important role in tumor progression, especially in tumor metastasis [64]. Integrins can interact with cell surface molecules and bind to the ECM to maintain cell viability and establish cell traction, thereby promoting cell migration and invasion [65]. During the PMN formation, the exosomal integrin αvβ3 can facilitate the specific organ metastasis by inducing vascular leakage, proinflammatory, and mediating intercellular crosstalk, etc [38]. It has been found that M2 macrophage-derived exosomal integrin αvβ3 can promote the lung cancer metastasis by activating the FAK signaling pathway [66]. Integrin αvβ3 can be present in prostate cancer-derived exosomes that can be transferred from tumorigenic cells to non-tumorigenic cells, thereby promoting tumor metastasis [67]. Moreover, integrin αvβ3 interacts with vitronectin to promote the colonization of ovarian cells in the liver by activating the cytoskeletal system [68]. Our findings are consistent with and extend previous observations in other tumor types. In contrast to these studies, our research specifically highlights the role of CRC cell-derived exosomal integrin αvβ3 in liver PMN formation. This study found that integrin αvβ3 expression was significantly upregulated in tumor tissues of CRC patients, suggesting a poor prognosis. In particular, integrin αvβ3 expression was significantly higher in the serum of patients with CRLM than patients without liver metastasis. In order to identify potential therapeutic markers of CRLM, we focused on exosomal integrin αvβ3. This study found that integrin αvβ3 was also highly expressed in the serum exosomes of patients with CRLM. The exosomal integrin αvβ3 may serve as a novel biomarker and therapeutic target for CRLM. Liver PMN formation contributes to the colonization of primary CRC cells in target organs. HSCs, which reside in the perisinusoidal space of the liver, are key cellular components of the liver PMN. Upon activation, HSCs can provide a suitable environment for tumor cell colonization in the liver by remodeling the ECM [68]. It has been found that HSCs can be activated by uptake of exosomal miR-181a-5p, which promotes the liver colonization of CRC cells through the secretion of CCL20 [69]. Many studies have shown that the activation of HSCs is associated with CRLM. Exosomes can deliver biomolecules to receptor cells in the liver PMN, which can induce phenotypic and functional changes in target cells, thereby promoting the liver PMN formation [20]. The current studies mainly focus on the effects of the exosomal integrin on target cells, but how the exosomal integrins target resident cells in PMN has not been elaborated. In this study, we first found that exosomal integrin αvβ3 can be taken up by HSCs via the immunofluorescence assay. Further exploring how the exosomal integrin αvβ3 activates HSCs, our mechanistic data revealed that the interaction between exosomal integrin αvβ3 and vitronectin on HSCs surface is crucial for initiating downstream FAK activation, which in turn drives the activation of the MEK-ERK cascade. Exosomes primarily exert their biological functions by delivering cargo through ligand-receptor interactions, membrane fusion, or endocytic mechanisms [36, 70]. Based on these established pathways, we propose two potential mechanisms by which exosomal integrin αvβ3 activates signaling cascades in HSCs: (1) Integrin αvβ3 on the exosomal membrane directly binds to vitronectin on the HSCs surface, triggering downstream signaling pathway activation; (2) Exosomes fuse with the HSCs membrane, releasing integrin αvβ3 into the intracellular space, where it interacts with vitronectin to activate downstream signaling. Notably, the upstream and downstream regulatory relationship between vitronectin and HSCs activation has not been fully elucidated. Future studies will focus on clarifying how exosomal integrin αvβ3 interacts with vitronectin, the temporal sequence and casual relationship between vitronectin expression and HSCs activation to address this knowledge gap. Many studies have confirmed that CAFs can promote the tumor cells proliferation and invasion by secreting a variety of biomolecules [71], and establish PMN that supports the colonization, growth and invasion of tumor cells in specific metastatic organs. The liver PMN plays a key role in the process of CRLM, especially through cell crosstalk between tumor cells and CAFs [72]. It has been found that exosomal HSPC111 can regulate lipid metabolism in CAFs by secreting CXCL5, thereby inducing EMT in CRC cells and ultimately promoting CRLM [73]. Another study found that the expression of CAFs markers in CRLM specimens is higher than that in non-metastatic liver specimens [72]. CAFs are key stromal cells in the tumor microenvironment, and a large amount of evidence shows that CAFs can participate in the PMN formation by promoting angiogenesis, suppressing immunity, releasing inflammatory factors, inducing EMT and other pathways [74, 75]. This study further found that CAFs can reshape liver PMN through EMT, angiogenesis and pro-inflammatory microenvironment, thereby promoting CRLM. This study also found that clinical studies of serum exosomal integrin αvβ3 expression in CRC patients provide strong evidence supporting its potential as a diagnostic indicator for CRLM. Given the current small sample size, future studies need to expand the cohort size for clinical validation to enhance the generalizability of these findings. In addition, KM survival analysis confirmed that the OS and DFS of patients with high expression of exosoaml integrin αvβ3 were shorter than those patients with low expression. At present, some integrin-targeted inhibitors have been used in the cancer treatment, highlighting integrins as promising therapeutic targets for malignant tumors [76]. In our study, treatment with cilengitide, an integrin inhibitor, significantly attenuated HSCs activation, further validating the central role of the integrin αvβ3-FAK axis in the liver PMN formation. Cilengitide-mediated inhibition of integrin αvβ3 can regulate stromal cell activation and angiogenesis [77]. Notably, cilengitide can simultaneously target integrin αvβ3 expressed on the surface of both tumor cells and stromal cells, thereby achieving potent blockade of tumor-stroma crosstalk. Specifically, cilengitide not only competitively binds to exosomal integrin αvβ3 to block its activating effect on HSCs but also inhibits integrin αvβ3 expressed on the cell membrane, thereby amplifying its inhibitory efficacy. These findings reveal a novel mechanism by which exosomes regulate CRLM and identify a promising potential target for CRLM inhibition. Moreover, exosomal integrin αvβ3 may serve as a valuable biomarker for CRLM. While our results are compelling, several limitations warrant discussion. First, the sample size for clinical correlation remains relatively small. Larger and multicenter studies are necessary to robustly validate exosomal integrin αvβ3 as a diagnostic marker for CRLM. Second, although we have delineated a clear pathway by which exosomal integrin αvβ3 drives HSCs activation via the vitronectin-FAK/MEK-ERK axis, the possibility of additional co-factors or parallel signaling cascades contributing to PMN formation cannot be excluded. Future studies should employ in vivo models and advanced proteomic analyses to explore these aspects further. In summary, our study not only elucidates a novel mechanism by which CRC-derived exosomal integrin αvβ3 promotes liver metastasis but also provides a strong rationale for the development of integrin-targeted therapies. By contextualizing our findings within the broader landscape of integrin-related research across other tumor types, we underscored the universal role of integrin-mediated signaling in tumor metastasis and highlighted exosomal integrin αvβ3 as a promising target for therapeutic intervention in CRLM. Conclusion This study revealed a novel mechanism by which CRC-derived exosomal integrin αvβ3 in CRLM. CRC-derived exosomal integrin αvβ3 specifically targeted vitronectin expressed on HSCs. The interaction activated the vitronectin-integrin αvβ3-FAK signaling axis, which in turn triggered the downstream MEK-ERK cascade to drive HSCs activation and their transformation into CAFs. These activated CAFs then reshaped the liver PMN through EMT, angiogenesis, and pro-inflammatory microenvironment. Collectively, these processes facilitated the colonization of CRC cells in the liver, ultimately promoting CRLM (Fig. 7). Our results supported the potential of exosomal integrin αvβ3 as both a biomarker and a therapeutic target for CRLM.Fig. 7Schematic diagram of the mechanism by which the CRC-derived exosomal integrin αvβ3 is involved in the liver PMN formation by inducing the CAFs formation. Exosomal integrin αvβ3 targets HSCs through vitronectin, thereby activating the FAK/MEK-ERK signaling pathway. This activation drives the transformation of HSCs into CAFs and facilitates the establishment of a liver PMN Schematic diagram of the mechanism by which the CRC-derived exosomal integrin αvβ3 is involved in the liver PMN formation by inducing the CAFs formation. Exosomal integrin αvβ3 targets HSCs through vitronectin, thereby activating the FAK/MEK-ERK signaling pathway. This activation drives the transformation of HSCs into CAFs and facilitates the establishment of a liver PMN Electronic Supplementary Material Below is the link to the electronic supplementary material. Supplementary Material 1 Supplementary Material 1