Twenty male Sprague-Dawley (SD) rats, aged 6–8 weeks and weighing 180∼220 g, were purchased from the Laboratory Animal Center of Sun Yat-sen University. The animal production license number is SCXK (Guangdong) 2021-0029. The animal use license number is SYXK (Guangdong) 2021-0112. All experimental procedures were performed in accordance with the Guidelines for the Care and Use of Laboratory Animals published by the US National Institutes of Health and approved by the Animal Care and Use Committee of the School of Life Sciences, Sun-Yat-sen University (Approval No. SYSU-IACUC-2023-000933, Approval Date: 2023-06-05).

SD rats were randomly divided into two groups (10/group) and received oral gavage treatment for 7 days: 1. Blank control group, normal diet, saline administration (5 mL/kg/d); 2. NXT group, normal diet, NXT administration (1,000 mg/kg/d). Three NXT doses of 250, 500, and 1,000 mg/kg/d were used in the preliminary study of our team, among which 1,000 mg/kg/d had the most significant inhibitory effect on endothelial dysfunction and oxidative stress, so the 1,000 mg/kg/d NXT dose was selected for subsequent gavage (Zhang W. J. et al., 2020). NXT (Med-drug permit no. Z20025001) was obtained from Shanxi Buchang Pharmaceutical Co. Ltd. (Xianyang, China). All identification of metabolites in NXT and quality control of the preparation strictly adhered to the standards specified under “Naoxintong Capsule” in the Chinese Pharmacopoeia (2020 Edition), which were implemented by the manufacturer. All NXT used in this study were from the same batch. Treatments were given and body weights were measured once a day. At the end of the experiment, blood was drawn from the abdominal aorta of SD rats to sterile centrifuge tubes. The 10 mL blood was extracted from each SD rat and was centrifuged at 8,000 g for 30 min at 4 °C. The sediment was discarded, and the serum was further purified by centrifuging at 10,000 g for 30 min using a Thermo Multifuge X1R centrifuge at 4 °C (Thermo Scientific, United States). A total of 110 mL serum was ultra-centrifuged at 4 °C using the SW28 rotors of an OptimaL-100XP centrifuge at 140,000 g for 90 min (Beckman, United States), and the circulating exosomes were dissolved in PBS for later identification (Wei et al., 2021).

The morphological characters of circulating exosomes were imaged by JEM-1400 Flash electron microscopy (JEOL, Japan) with phosphotungstic acid staining. The particle size distribution of circulating exosomes was analyzed by NanoSight NS300 (Malvern Panalytical, United Kingdom). Expressions of exosome surface markers (ALIX, CD9, and TSG101) and negative control marker (β-actin) were detected by Western blot. Concentrations of circulating exosomes were measured by BCA kit (P0012, Beyotime, China). The circulating exosome PBS solution was adjusted to 1 mg/mL based on protein concentration, aliquoted, and stored at −80 °C for subsequent use.

HMEC-1 (CRL-3243) and HUVEC (PCS-100-013) were purchased from American Tissue Culture Collection (ATCC, United States). Both cell types were cultured in cell incubators at 5% CO₂ and 37 °C using endothelial cell medium supplemented with 5% exosome-free fetal bovine serum, 1% penicillin/streptomycin, and 1% endothelial cell growth supplement (1001, ScienCell, Carlsbad, United States). The cell culture medium was replaced every 2 days, and cells reaching 80% density were passaged at a ratio of 1:3 density. HMEC-1 and HUVEC between passages 3 and 8 were used in this study.

Circulating exosomes were labeled with PKH26 (MINI26-1KT, Sigma, United States). The 400 μL PKH26 (2 μM) labelling solution was mixed with 200 μL circulating exosomes for 10 min, then 1.2 mL of 1% bovine serum albumin was added and mixed, followed by centrifugation at 100,000 g for 1 h to remove PKH26 labelling solution. The exosome sediment was then resuspended in PBS. HMEC-1 and HUVEC were stained with 50 μL Dio (HY-D0969, MCE, United States) and Hoechst 33342 (S0485, Selleck, United States). Subsequently, the PKH26-labeled circulating exosomes (5 μg/mL protein) were added into 12-well plates pre-inoculated with HMEC-1 and HUVEC for 2 h incubation. The cells were then fixed with 4% paraformaldehyde. Finally, the uptakes of circulating exosomes were visualized using an LSM 880 laser confocal microscope (Zeiss, Germany).

The CCK-8 assay (CK04, Dojindo, Japan) was used to test cell viability. Cells in logarithmic phase were seeded in 96-well plates (3 × 10³ cells per well) and cultured in ECM complete medium for 12 h to allow for cell adhesion. Subsequently, the medium was replaced with ECM complete medium containing 25, 50, 100, 200, 400, and 800 ng/mL LPS for 24 h. Following incubation, 10 μL CCK-8 reagent was added to each well and incubated at 37 °C for 1 h. Cell viability was determined by measuring the absorbance at 450 nm using a microplate detector (BioTek, United States). After the concentration of LPS applied to the model group was selected, cells were inoculated as described above. After cell adhesion, the medium was replaced by the ECM complete medium containing 200 ng/mL LPS, 200 ng/mL LPS + 2.5 μg/mL Ctl-Exo, 200 ng/mL LPS + 5 μg/mL Ctl-Exo, 200 ng/mL LPS + 10 μg/mL Ctl-Exo, or 200 ng/mL LPS + 2.5 μg/mL NXT-Exo, 200 ng/mL LPS + 5 μg/mL NXT-Exo, 200 ng/mL LPS + 10 μg/mL NXT-Exo for 24 h. Following treatment, cells were incubated with 10 μL CCK-8 reagent for 1 h at 37 °C, and the absorbance value at 450 nm was measured using a microplate detector.

The LDH concentrations of HMEC-1 and HUVEC were detected with the LDH kit (A020-2-2, Nanjingjiancheng, China) according to the manufacturer’s instructions. In the test hole, 16 μL sample solution, 20 μL matrix buffer, and 4 μL Coenzyme I solution were added successively and mixed, and the temperature bath was 37 °C for 15 min. Then 20 μL 2, 4-dinitrophenylhydrazine were added and bathed at 37 °C for 15 min, and finally 200 μL 0.4 mol/L NaOH was mixed. The absorbance value at 440 nm was measured by microplate reader.

Extracellular NO secretion levels of HMEC-1 and HUVEC were detected with the NO kit (A013-2-1, Nanjingjiancheng, China). According to the manufacturer’s instructions, 100 μL cell supernatants were collected and treated with 200 μL reagent 1 and 100 μL reagent 2 for 10 min. Following this, the mixture was centrifuged at 2,000 rcf for 15 min, and the resulting supernatants were collected. Then 160 μL supernatants were added to 80 μL chromogenic agents and allowed to react for 15 min. Subsequently, the absorbance value at 550 nm were detected by microplate reader.

The expression levels of IL-1β (SEA563Hu, Cloud-clone, China), IL-6 (SEA079Hu, Cloud-clone, China), IL-8 (SEA080Hu, Cloud-clone, China), VCAM-1 (SEKH-0055, Solarbio, China), ICAM-1 (SEKH-0053, Solarbio, China), eNOS (CSB-E08322h, Cusabio, China), and ET-1 (CEA482Hu, Cloud-clone, China) were measured by quantitative ELISA kits according to the kit protocol. First, 100 μL cell supernatants from HMEC-1 and HUVEC and 100 μL standards of different concentrations were added to the 96-well ELISA plate, and incubated at 37 °C for 1 h. Subsequently, the contents were discarded, and the wells were treated with the 100 μL working solution and 90 μL TMB substrate solution for the specified reaction time. Finally, 50 μL termination solutions were added and the OD450 values were immediately measured by microplate reader.

Apoptosis rates of HMEC-1 and HUVEC were detected with Annexin V, FITC apoptosis kit (AD10, Dojindo, Japan). Cells were centrifuged at 1,000 rcf for 3 min, and supernatants were discarded. Subsequently, cells were re-suspended in 500 μL 1 × Annexin V binding solution, and 100 μL suspension was incubated with 5 μL Annexin V-FITC and PI solution at 37 °C for 15 min away from light. Following this, 400 μL 1× Annexin V binding solution was added, and the samples were examined by flow cytometry. Unstained cells were used as the negative control.

Cells and circulating exosomes were lysed with ice-cold RIPA lysis buffer (P0013B, Beyotime, China). Samples (10 μg) were loaded onto 10% sodium dodecyl sulfate/polyacrylamide gel electrophoresis (1610173, Bio-Rad, United States) and subsequently transferred to polyvinylidene fluoride membranes. After transfer, the membranes were blocked with sealing liquid and incubated for 2 h. Following blocking, membranes were incubated with primary antibodies against β-actin (dilution 1:1,000, ab8227, Abcam), ALIX (dilution 1:1,000, ab275377, Abcam), TSG101 (dilution 1:1,000, ab125011, Abcam), CD9 (dilution 1:1,000, ab236630, Abcam), Bax (dilution 1:1,000, ab32503, Abcam), Bcl-2 (dilution 1:1,000, ab32124, Abcam), Cleaved caspase 3 (dilution 1:500, ab32042, Abcam), TLR4 (dilution 1:1,000, TA382568, OriGene), TRIF (dilution 1:1,000, TA382523, OriGene), NF-κB p65 (dilution 1:1,000, TA385159, OriGene), Phospho-NF-κB p-p65 (dilution 1:1,000, TA380848, OriGene), STC1 (dilution 1:5000, ab229477, Abcam) at 4 °C overnight. Subsequently, the membranes were incubated with the secondary antibody horseradish peroxidase-labeled goat anti-rabbit immunoglobulin G (dilution 1:5,000, ab6721, Abcam). Immunoblots were visualized using the Tanon 5200 chemiluminescence imaging system (Tanon, China) and analyzed with the ImageJ software.

The production levels of ROS in HMEC-1 and HUVEC were determined by fluorometric assay with DCFH-DA as a probe for the presence of ROS (R252, Dojindo, Japan). Cells in logarithmic phase were seeded in 96-well black plates (3 × 10³ cells per well), and cultured in ECM complete medium for 12 h. After treatment, supernatants were removed and 100 μL DCFH-DA working solution was added per well. Samples were then incubated at 37 °C for 30 min. Subsequently, the DCFH-DA working solution was removed, and the cells were washed twice with PBS. Cells were detected under Spark fluorescence microplate reader (Tecan Spark, Switzerland).

According to the instructions of the MDA kit (M496, Dojindo, Japan), cell suspensions were collected and centrifuged, mixed with 100 μL Antioxidant PBS solution, and transferred into the labeled sample tube. The MDA standard solution with a concentration of 10 μmol/L was diluted in equal proportion to different concentrations for the preparation of the standard curve. Then, 100 μL lysis buffer and 250 μL working solution were added to sample tubes and standard tubes. After mixing, the tubes were heated in a metal bath at 95 °C for 15 min, immediately bathed in an ice bath for 5 min, cooled to room temperature, centrifuged at 10,000 g for 10 min, and 100 μL supernatant was taken and added to a black 96-well plate. The fluorescence intensity was detected with a fluorescent enzyme marker, and the parameters were set as follows: excitation wavelength was 540 nm, emission wavelength was 590 nm. The MDA concentration was calculated according to the MDA standard curve.

After cell treatment, 1 mL TRIzol reagent was added to each hole of the 6-well plate and cracked on ice. The cell lysate was collected and stored in the frozen storage tube and stored in the refrigerator at −80 °C for sample testing. The cDNA libraries were sequenced on the Illumina sequencing platform by Genedenovo Biotechnology Co., Ltd. (Guangzhou, China). In brief, the testing process begins with sample testing, then cDNA library construction, library quality control, and finally sequencing and data processing. Ctl-Exo and NXT-Exo samples stored at −80 °C were sent for miRNA sequencing (LC-Bio Technologies, Hangzhou, China). Similarly, the first step was library construction, then sequencing, and finally data analysis. The mRNA and miRNA sequencing data of HMEC-1 have been deposited in the NCBI Gene Expression Omnibus (GEO) database and are accessible through GEO Series accession number GSE235049 and GSE298553. The threshold value for selection of DEGs and DEMs was p-value <0.05 and fold change (FC) ≥ 2 or ≤0.5, and the results were filtered using the False Discovery Rate (FDR) < 0.05. KEGG classifications were used to analyse DEGs.

Total RNA was extracted with EZ-press RNA Purification Kit (B0004D, EZBioscience, United States) from HMEC-1 under the guidance of the manufacturer’s protocol. Total RNA was isolated from circulating exosomes using TRIzol reagent (15596018CN, Thermo Fisher Scientific, United States). The concentration of the isolated RNA samples was measured by Nanodrop 2000 (Thermo Fisher Scientific, United States), and cDNA was obtained by reverse transcription of cells RNA with StarScript Pro reverse transcription premix (A240, Genstar, China) and by reverse transcription of circulating exosomes RNA with Synthesis of miRNA first strand cDNA (B532451-0020, Sangon Biotech, China). RT-qPCR was then performed using GoTaq qPCR Master Mix (A6002, Promega, United States), following the manufacturer’s instruction. Reaction procedures were 95 °C for 3 min, then 45 cycles at 95 °C for 10 s and 60 °C for 30 s. Primers were synthesized by Sangon (Sangon Biotech, China). The reverse primers of these four miRNAs and the primers of U6 were universal and provided in the miRNA first strand cDNA kit. Relative levels of mRNA and miRNA were normalized to β-actin and U6, and the relative expression levels of mRNA and miRNA were calculated using the 2^−ΔΔCT method. The primer sequences of mRNAs are shown in Table 1.

The DEMs between Ctl-Exo and NXT-Exo groups were screened and summarized. Target genes were predicted by Miranda (v3.3a) and TargetScan (Version: 7.0) methods. TargetScan score ≥50, Miranda Energy < −10. The intersection of target gene prediction results obtained by the two methods was taken as the result of miRNAs target gene prediction. The predicted miRNAs target genes were then intersected with the DEGs in the RNA-seq of HMEC-1 to determine the differentially targeted genes of miRNAs in NXT-Exo’s anti-endothelial injury effect. CytoScape 3.10.1 software was used to map miRNA-mRNA networks to analyze and screen key miRNAs.

All the miRNA mimics, miRNA inhibitors and si-STC1 used in the study were synthesized by Sangon (Sangon Biotech, China). miRNA and si-RNA transfection were performed using the RNA transfection reagent (E607402, Sangon Biotech, China) and followed the manufacturer’s direction. The final concentration of 20 μM was prepared by adding DEPC water into the miRNA and si-RNA tube. The transfection reagent was prepared by taking 380 μL ECM complete medium from tube A, adding 24 μL RNATransMate reagent and mixing well. Tube B was mixed 20 μL miRNA inhibitor, 10 μL miRNA mimic or 20 μL si-STC1 with 200 μL ECM. Tube A and tube B were mixed to make transfection mixture, and the mixture was left at room temperature for 10 min miRNA inhibitor/RNA TransMate, miRNA mimic/RNA TransMate or si-STC1/RNA TransMate complexes were formed. HMEC-1 were inoculated into 24-well plates in advance. When the cell density reached 80%, 50 μL complexes and 450 μL ECM were added into each well. The final concentration of miRNA inhibitor and si-STC1 was 50 nM, and the final concentration of miRNA mimic was 25 nM. The 24-well plates were transfected in a 37 °C incubator for 6 h, and the liquid was discarded. The group molding was performed and the drug was administered for 24 h. The preparation methods of miRNA inhibitor-NC/RNA TransMate, miRNA mimic-NC/RNA TransMate, and si-STC1-NC/RNA TransMate complexes were the same as above (Tong et al., 2023). The primer sequences of miRNAs are shown in Table 2.

Scientific illustrations were created with BioRender. Experimental data were analyzed using Graphpad Prism 9.3.1 and expressed as the mean ± SD. The significance of the difference between two groups was assessed by Student’s t-test. Multiple group differences were compared using one-way analysis of variance (ANOVA). p < 0.05 was considered statistically significant.

To monitor the effects of NXT on SD rats, we daily administered SD rats and recorded their body weights. Compared with the blank control group orally administered with saline, the NXT group had no obvious weight changes, confirming that NXT did not affect the healthy growth of SD rats (Figure 1A). We then extracted NXT-Exo and Ctl-Exo, characterizing them using transmission electron microscopy (TEM), nanoparticle tracking analysis (NTA), and Western blot analysis. TEM indicated that the exosomes exhibited a typical spherical bilayer membrane structure with diameters of approximately 100 ∼ 150 nm (Figure 1B), while NTA showed that the majority of exosome particles diameter <200 nm. The results of both experiments were consistent (Figure 1C). Western blot results confirmed the presence of exosomal surface markers ALIX, CD9, and TSG101 (Figure 1D). Crucially, both NXT-Exo and Ctl-Exo exhibited typical exosome morphology and diameter distribution, while the protein concentration of NXT-Exo was significantly higher than that of Ctl-Exo, as measured by BCA (Figure 1E). This suggested that NXT stimulated the generation of circulating exosomes. Furthermore, we observed the internalization of PKH26-stained circulating exosomes by both HMEC-1, demonstrating the ability of circulating exosomes to enter VECs and potentially deliver signals (Figure 1F).

To investigate the protective effects of NXT-Exo against LPS-induced injury in HMEC-1, cells were exposed to increasing gradient doses of LPS for 24 h. An initial dose-response experiment revealed a decrease in cell viability in an LPS dose-dependent manner, with significant inhibition at concentrations ≥200 ng/mL (Figure 2A). We therefore selected 200 ng/mL LPS for subsequent experiments. In order to verify whether circulating exosomes can reduce the VECs injury induced by LPS, HMEC-1 were treated with Ctl-Exo or NXT-Exo in the presence of LPS. While Ctl-Exo had no significant effect on cell viability, NXT-Exo at 5 and 10 μg/mL significantly improved the cell viability in HMEC-1 compared to LPS treatment (Figure 2B). Considering the limited and valuable extractions of circulating exosomes from SD rats, a single dose of 5 μg/mL Ctl-Exo and NXT-Exo was chosen for subsequent cell experiments in order to comply with the “3R principle” (i.e., Replacement, Reduction, and Refinement) of experimental animal ethics and minimize exosome wastage.

Regulating the dynamic balance of relaxation and contraction to maintain homeostasis is an important function of VECs (Gupta et al., 2020). To investigate the effect of NXT-Exo on endothelial tension, eNOS, NO, and ET-1 levels were measured. The results showed that compared with the NC group, the expression of eNOS and NO in the LPS group was downregulated, and the expression of ET-1 was upregulated. The NO elevation may be associated with enhanced eNOS activity. NXT-Exo significantly reversed the LPS-induced downregulation of NO (Figure 2C) and eNOS (Figure 2D) and upregulation of ET-1 (Figure 2E), whereas Ctl-Exo had no significant effect.

To further evaluate the protective effect of NXT-Exo against LPS-induced apoptosis, we measured cell apoptosis rates using flow cytometry. Results indicated that NXT-Exo significantly reduced LPS-induced apoptosis (Figure 2F). Additionally, Western blot analysis revealed that NXT-Exo markedly attenuated LPS-induced upregulation of Bax and Cleaved caspase 3 and downregulation of Bcl-2 in HMEC-1 (Figure 2G).

To assess the extent of cell damage, we employed the LDH assay and observed that, compared with the LPS group, Ctl-Exo treatment exhibited no significant change in LDH level, whereas NXT-Exo treatment significantly reduced LDH level (Figure 3A). These results suggested that NXT-Exo could possess the capability to mitigate LPS-induced VECs injury. Given that inflammation-mediated VECs damage greatly increases the risk of cardiovascular and cerebrovascular diseases, we examined inflammatory factors using ELISA assay. Our findings demonstrated that the secretion levels of IL-1β, IL-6, IL-8, and adhesion molecules VCAM-1, ICAM-1 were notably increased following LPS challenge. However, treatment with NXT-Exo effectively inhibited the LPS-induced elevation of these factors (Figures 3B–F). Moreover, considering that LPS-induced ROS generation is a major contributor to cytotoxicity, we assessed the ROS fluorescence intensity and MDA concentration. Our results revealed a significant increase in ROS and MDA levels in the LPS group, while NXT-Exo treatment considerably alleviated these markers (Figures 3G,H).

To identify the molecular mechanisms underlying NXT-Exo’s protective effects against LPS-induced VECs injury, we performed RNA-seq analysis on HMEC-1, using FC ≥ 2 or ≤0.5, FDR < 0.05 and p-value <0.05 as the screening criteria. We analyzed DEGs between the following groups: NC vs. LPS, LPS vs. LPS + NXT-Exo, NC vs. LPS + Ctl-Exo, and LPS + Ctl-Exo vs. LPS + NXT-Exo. As anticipated, numerous genes were differentially expressed in LPS compared to NC, with more genes were upregulated in LPS. Interestingly, this imbalance was mitigated in the LPS + NXT-Exo group, suggesting a protective effect of NXT-Exo against LPS injury. In contrast, the LPS + Ctl-Exo group had little effect in reversing the LPS-induced gene upregulation (Figure 4A). We then conducted KEGG enrichment analysis of DEGs between groups, highlighting the top pathways involved in DNA replication, lysosome, cell cycle, protein processing in endoplasmic reticulum and apoptosis (Figure 4B). These pathways mirrored the potential mechanisms of NXT-Exo’s protection against LPS-induced cell injury. Based on the previous findings and relevant literature, we selected ten representative genes related to vascular endothelial injury, including CDH5, BAX, BCL2, ATG13, ATG14, IL-6, IL-8, IL6ST, TNFRSF9, and IGSF8 (Figure 4C), and confirmed their differential expression by RT-qPCR (Figure 4D). The RT-qPCR validation of these key genes strengthened our confidence in the RNA-seq findings and provided specific targets for further mechanistic investigation.

In vivo, LPS can be recognized and bound by lipopolysaccharide binding protein (LBP) due to its high affinity with lipid A structure. Subsequently, LPS forms a complex with CD14 under the action of lipid shuttle and presentation by LBP, resulting the formation of LPS-LBP-CD14 complex (Mattingly et al., 2022). This complex is recognized by TLR4 receptor on the VECs membrane through the PAPM mode, leading to vascular endothelial inflammation and subsequent oxidative and apoptotic damage (Marongiu et al., 2019). Studies have shown that active metabolites of natural herbal extracts (Fu et al., 2020), Chinese herbal compound preparations (Miao et al., 2022), intestinal microbiota (Li et al., 2024), and specific nucleotides (Cai et al., 2022) inhibited LPS-induced inflammation by regulating TLR4/TRIF/NF-κB or TLR4/NF-κB pathways. Based on these findings, we began to elucidate the pathway through which NXT-Exo mitigated cell injury. Consistent with this pathway, our RNA-seq analysis revealed that DEGs were related to inflammation and apoptosis, and KEGG enrichment analysis highlighted biological processes including cell cycle, lysosome, and apoptosis, which were closely related to inflammation. These results suggested that NXT-Exo may be involved in the TLR4/TRIF/NF-κB or TLR4/NF-κB pathway in ameliorating LPS-induced cell injury. To test our hypothesis, we examined the expression of TLR4, TRIF, and p65 using RNA-seq and confirmed them by RT-qPCR. Remarkably, their expression levels were significantly increased in the LPS group, but decreased after NXT-Exo treatment (Figure 4E). Western blot assay further validated these findings at the protein level, showing increased expression of TLR4, TRIF, and p-p65/p65 ratio in the LPS group, all of which were reversed by NXT-Exo treatment. Therefore, the addition of the NF-κB activator PMA attenuated the inhibitory effects of LPS + NXT-Exo on TLR4, TRIF, and p-p65/p65 (Figure 4F), suggesting that NXT-Exo may exert NXT pharmacological effects by suppressing the TLR4/TRIF/NF-κB signaling pathway.

To investigate whether NXT-Exo can inhibit LPS-induced HMEC-1 damage through miRNA, we performed miRNA-omics analysis of Ctl-Exo and NXT-Exo, using FC ≥ 2 or ≤0.5, FDR <0.05 and p-value <0.05 as the screening criteria. A total of 12 DEMs were screened out, hierarchical clustering heatmap shows the 12 increased and decreased miRNAs (Figure 5A), including eight upregulated miRNAs and four downregulated miRNAs (Figure 5B). Miranda (v3.3a) and Target Scan (Version: 7.0) were used to predict the target genes of these DEMs in Ctl-Exo and NXT-Exo, with screening criteria of Target Scan score ≥50 and Miranda energy < −10. The intersection of the two sets revealed 7,707 target genes. In order to ensure the reliability, the predicted target genes of miRNAs were intersected with the DEGs of RNA-seq, resulting in 2,384 overlapped genes (Figure 5C). The miRNA-mRNA network map was constructed based on these overlapped genes, which showed that all the 12 miRNAs could target the overlapped genes (Figure 5D).

Through literature analysis, we found that six out of the 12 DEMs have been reported to be associated with vascular endothelial injury or cardiovascular and cerebrovascular diseases. By comparing the target genes mentioned above with the overlapped genes, we identified miR-202-5p, miR-382-5p, miR-1-3p, and miR-802-5p as core miRNAs (Figure 5E) and verified their differential expression by RT-qPCR. The result of RT-qPCR was consistent with miRNA-omics analysis (Figure 5F).

To further study the protective effects of miRNAs, we inhibited or increased the core miRNAs expression in HMEC-1 and investigated three aspects of cell damage, including LDH release, IL-6 secretion and ROS production. Compared with the LPS + NXT-Exo group, the LPS + NXT-Exo + miR-382-5p mimic group showed a significant increase in LDH release, IL-6 secretion, and ROS production. Additionally, the LPS + NXT-Exo + miR-1-3p mimic group showed an increase in IL-6 secretion. In contrast, the LPS + NXT-Exo + miR-202-5p inhibitor group and LPS + NXT-Exo + miR-802-5p inhibitor group showed no significant changes. These results suggested that miR-382-5p played a key role in the protective effects of NXT-Exo against LPS-induced inflammatory oxidative damage in HMEC-1 (Figures 6A–C).

Based on our integrative studies above, we hypothesized that STC1 may be regulated by miR-382-5p and contribute to NXT-Exo’s protection against LPS-induced HMEC-1 damage. Stanniocalcin-1 (STC1), a 35 kD homologous disaccharide protein hormone, is expressed in various cells such as neurons, megakaryocytes, and endothelial cells, possessing antioxidant and anti-inflammatory properties (Shi et al., 2018). STC1 has been confirmed as a target of miR-382-5p, and it is involved in regulating cardiomyocyte apoptosis following acute myocardial infarction. To verify whether miR-382-5p affects the expression of STC1, we transfected HMEC-1 with miR-382-5p mimic or miR-382-5p mimic-NC, and detected the expression of STC1 using RT-qPCR. The results showed that miR-382-5p mimic significantly inhibited the expression of STC1, suggesting STC1 may be the target of miR-382-5p (Figure 7A).

Additionally, we co-treated HMEC-1 with NXT-Exo and either miR-382-5p mimic or miR-382-5p mimic-NC. Compared with the LPS group, NXT-Exo treatment upregulated STC1 expression. Transfection with miR-382-5p mimic in the presence of NXT-Exo led to downregulation of STC1 expression compared with the LPS + NXT-Exo group, indicating that miR-382-5p could reverse the promoting effect of NXT-Exo on STC1 expression. This trend of STC1 expression was consistent with the results of RNA-seq (Figure 7B). In conclusion, miR-382-5p downregulated the expression of STC1, suggesting that the protective effects of NXT-Exo against LPS-induced HMEC-1 damage involved the modulation of STC1 through miR-382-5p.

As miR-382-5p could inhibit the expression of STC1, we further explored whether STC1 was responsible for the protective effect of NXT-Exo. We transfected HMEC-1 with si-STC1 and verified STC1 expression by RT-qPCR. The expression level of STC1 mRNA significantly decreased after transfection with si-STC1 compared to si-STC1-NC, indicating successful transfection and effective inhibition of STC1 expression (Figure 7C). Next, we examined the effect of STC1 on various cell damage markers, including LDH, IL-6, IL-8, ROS, and MDA. The results showed that si-STC1 counteracted NXT-Exo’s effect on the downregulation of these markers (Figures 7D–H), suggesting that STC1 exerted anti-inflammatory and antioxidant effects, inhibiting cell damage. Thus, STC1 might be a key molecule mediating the protective effects of NXT-Exo against LPS-induced cell injury.

To determine the protective effects of miR-382-5p and STC1 on VECs apoptosis, we transfected HMEC-1 with si-STC1 and miR-382-5p mimic. Western blot results revealed that transfection with miR-382-5p mimic resulted in upregulated levels of pro-apoptotic markers Bax and Cleaved caspase 3, and downregulated levels of anti-apoptotic markers Bcl-2 and STC1. Furthermore, co-transfection with miR-382-5p mimic and si-STC1 significantly increased the expression levels of Bax and Cleaved caspase 3, while decreasing the expression levels of Bcl-2 and STC1 compared to cells transfected with miR-382-5p mimic alone (Figure 7I). These results indicated that miR-382-5p promoted LPS-induced apoptosis of HMEC-1 by down-regulating STC1, emphasizing the role of the miR-382-5p/STC1 axis in VECs apoptosis.

To validate the effects of NXT-Exo on vascular endothelial cells, we employed primary HUVEC for supplementary verification. Initial fluorescence experiments confirmed HUVEC uptake of PKH26-labeled circulating exosomes (Figure 8A). Subsequently, establishment of LPS and exosome concentration gradients revealed that LPS dose-dependently reduced HUVEC viability, with significant inhibition observed at concentrations ≥200 ng/mL, thus establishing 200 ng/mL as the optimal LPS concentration for subsequent experiments (Figure 8B). NXT-Exo at 5 and 10 μg/mL significantly enhanced cell survival rates in LPS-treated HUVEC, whereas Ctl-Exo showed no protective effects. The maximal protective effect of NXT-Exo was observed at 5 μg/mL, so 5 μg/mL was selected as the concentration for both Ctl-Exo and NXT-Exo in subsequent experiments (Figure 8C). We then examined NXT-Exo’s impacts on endothelial dysfunction and apoptosis. NXT-Exo effectively reversed LPS-induced downregulation of NO (Figure 8D) and eNOS (Figure 8E), while mitigating ET-1 upregulation in HUVEC (Figure 8F). Furthermore, NXT-Exo reduced LPS-induced apoptosis (Figure 8G). Western blot analysis demonstrated that NXT-Exo significantly attenuated LPS-induced upregulation of Bax and Cleaved caspase 3, along with restoration of Bcl-2 expression in HUVEC (Figure 8H), collectively indicating NXT-Exo’s protective role against LPS-induced endothelial dysfunction and apoptosis. Moreover, we investigated NXT-Exo’s effects on injury and inflammatory oxidative responses post-LPS challenge. The results showed that LPS treatment elevated LDH, IL-6, IL-1β, IL-8, VCAM-1, and ICAM-1 levels, which were effectively counteracted by NXT-Exo intervention (Figures 9A–F). Oxidative stress evaluation revealed that NXT-Exo significantly suppressed LPS-induced increases in ROS production and MDA activity in HUVEC (Figures 9G,H), suggesting NXT-Exo’s capacity to restore anti-inflammatory and antioxidant activities following LPS-induced damage. These findings aligned with the HMEC-1 experimental outcomes, further suggesting that circulating exosomes induced by NXT may confer partial protection against endothelial injury.