A safe and anti-inflammatory plant-derived nanovesicle platform for targeted delivery in acute lung injury

Abstract

Acute lung injury (ALI) and its more severe form, acute respiratory distress syndrome (ARDS), are life-threatening pulmonary disorders with extremely high mortality rates, for which effective and safe therapeutic strategies remain limited. The development of targeted and biocompatible drug delivery systems is urgently needed to control pulmonary inflammatory cascades while minimizing systemic toxicity. Plant-derived extracellular vesicles offer a naturally safe and anti-inflammatory platform for therapeutic delivery. Ginsenoside Rb1 (GRb1), a major bioactive compound from ginseng, possesses potent anti-inflammatory and anti-apoptotic properties, whereas lemon-derived EVs (LEVs) exhibit intrinsic antioxidant and anti-inflammatory effects. Here, we engineered a multifunctional, biocompatible drug delivery platform, GRb1@LEVs-cRGD, in which ginsenoside Rb1 is incorporated into and fused with LEVs to form hybrid bio-nanovesicles, while the vesicle surface is functionalized with cyclic RGD (cRGD) peptides to target integrin αvβ3 highly expressed in inflamed pulmonary tissues, thereby enhancing site-specific delivery. In vitro and in vivo studies confirmed that GRb1@LEVs-cRGD effectively inhibited M1 macrophage polarization, suppressed inflammatory cascades, and preserved epithelial-endothelial integrity. Furthermore, exogenous cholesterol loading improved vesicle stability, maintained the pH gradient, and enhanced the loading efficiency of tigecycline and vancomycin by six-fold. In murine models of bacterial pneumonia induced by carbapenem-resistant Klebsiella pneumoniae and methicillin-resistant Staphylococcus aureus, antibiotic-loaded GRb1@LEVs-cRGD efficiently accumulated at infection sites and exhibited synergistic anti-inflammatory and bactericidal effects. Overall, this study demonstrates that GRb1@LEVs-cRGD is a safe, targeted, and multifunctional therapeutic platform with significant potential for ALI/ARDS treatment.

Graphical abstract

Highlights

• •A safe and anti-inflammatory plant-derived nanovesicle platform was created via membrane fusion.
• •This platform possesses excellent targeting capability toward inflamed lung tissues, thereby significantly enhancing drug accumulation at specific sites and minimizing off-target effects.
• •This platform exhibits excellent biocompatibility and well-defined core therapeutic efficacy, directly targeting the key pathological mechanisms of acute lung injury/acute respiratory distress syndrome (ALI/ARDS).
• •Through exogenous cholesterol loading, this platform achieves carrier optimization and upgrading, enhances vesicle stability, and improves the loading efficiency of antibiotics such as tigecycline and vancomycin by 6-fold.
• •This platform can achieve a synergistic anti-inflammatory and bactericidal effect in bacterial pneumonia models induced by multidrug-resistant bacteria.

1. Introduction

Acute lung injury (ALI) and its more severe form, acute respiratory distress syndrome (ARDS), are common life-threatening pulmonary disorders primarily associated with acute and severe pulmonary inflammation [1]. Globally, ALI/ARDS affects millions of patients annually, accounting for 10% of intensive care unit (ICU) admissions and representing a significant contributor to morbidity and mortality among critically ill individuals [2]. However, current therapeutic approaches remain focused on post-onset symptomatic and supportive care, including mechanical ventilation, prophylactic or therapeutic administration of antibiotics, and restriction of fluid accumulation [3] [4]. Despite such interventions, ALI/ARDS typically progresses to severe respiratory failure and death, with in-hospital mortality rates ranging from 38% to 46% [5,6]. Survivors also frequently develop long-term physical, psychological, and/or cognitive impairments [3,7]. Thus, there is an urgent need for therapeutic strategies to prevent or control pulmonary inflammatory cascades, aiming to reduce mortality and long-term morbidity in ALI/ARDS.

The pathogenesis of ALI involves damage to vascular endothelial and alveolar epithelial cells, leading to increased alveolar-capillary permeability and reduced alveolar surfactant [8,9]. Following the invasion of lung tissues by endotoxins, microorganisms, and other pathogens, a large number of inflammatory factors are released, which activate effector cells such as alveolar macrophages, epithelial cells, and endothelial cells, triggering an uncontrolled release of inflammatory cascades [10]. Thus, the rapid clearance of pulmonary pathogenic microorganisms and restoration of host homeostasis are crucial for the development of therapies targeting ALI/ARDS. Furthermore, due to the unique anatomical structure and immune microenvironment of the lungs, there is an urgent need to develop functional nanodrug delivery platforms with enhanced in vivo targeting, improved antibiotic-loading capacity, and superior biocompatibility [8,11]^, [12]^, [13].

Plant-derived extracellular vesicles (PDEVs), as edible components, have emerged as novel nanoscale drug delivery carriers in recent years, owing to their advantages of high yield, low cost, and excellent biosafety [14]. Moreover, the lipids, proteins, and nucleic acids inherently carried by PDEVs exhibit inherent anti-inflammatory, antioxidant, antibacterial, and antitumor properties, endowing them with extensive therapeutic value [[15], [16], [17]]. To date, exosomes derived from plants like lemon, grapefruit, and turmeric have been reported to exhibit significant efficacy in treating inflammation, infections, tumors, and other diseases. Among them, lemon-derived extracellular vesicles (LEVs) stand out for their superior performance in inhibiting inflammation and regulating immunity [[18], [19], [20]]. In addition to their intrinsic therapeutic capacity derived from innate bioactive metabolites, PDEVs are also suitable as vesicular carriers for drug delivery. Three registered clinical trials of PDEVs (ClinicalTrials.gov: NCT01668849, NCT03493984, NCT01294072) have investigated their feasibility as both active ingredients and drug delivery carriers [14]. However, despite their excellent biocompatibility and intrinsic therapeutic properties, the effective targeting of PDEVs to specific inflammatory sites and their capacity to deliver anti-inflammatory drugs with high efficiency remain challenging, limiting their clinical translation.

Ginseng ranks among the most renowned and valuable herbal medicines in East Asia [21], and ginsenoside Rb1 (GRb1) has been identified as one of its key bioactive components [22]. Studies have demonstrated that GRb1 exerts multiple beneficial effects on the human body, including inhibiting apoptosis, exerting anti-inflammatory activity, and alleviating oxidative stress [23,24]. Notably, GRb1 has been shown to suppress inflammation and oxidative stress-induced apoptosis in the heart, liver, and brain. Additionally, the structure of GRb1 contains hydrophobic triterpenoid or steroid aglycones and hydrophilic sugar side chains, enabling it to form micelles in aqueous solutions [[25], [26], [27]]. Therefore, taking advantage of the lipid bilayer structure of LEVs, we incorporated GRb1 into LEVs, resulting in the formation of hybrid nanovesicles (GRb1@LEVs). GRb1@LEVs integrates the anti-inflammatory and anti-apoptotic properties of LEVs and GRb1, while also possessing the ability to deliver hydrophobic drugs such as antibiotics as a carrier. This nanomedicine delivery vector is likely to exert a synergistic effect in the treatment of ALI/ARDS, achieving complementary therapeutic efficacy.

To maximize the therapeutic efficacy of bioinspired nanovesicles for ALI/ARDS while minimizing off-target effects, surface engineering is crucial [28]. PDEVs exhibit bio-membrane properties similar to those of mammalian-derived nanovesicles, thereby offering the possibility of surface customization, which can broaden the range of desired targeting capabilities [20]. Specific bioactive molecules are immobilized on the surface of PDEV-based nanoplatforms to develop displays with enhanced compatibility with cell surface receptors. Among these, the most common strategy for immobilizing specific active molecules on PDEVs surfaces is the incubation method, which involves mechanisms such as hydrophobic interactions, diffusion processes, and electrostatic interactions without compromising the integrity or functionality of the vesicles [29]. For instance, ginger-derived GDEVs are conjugated with FA-PEG-Chol for further targeted therapy of rheumatoid arthritis (RA) against activated macrophages [30]. Using DSPE-PEG-HA conjugates, hyaluronic acid (HA) is modified on the surface of red cabbage-derived extracellular vesicles for targeted treatment of inflammatory bowel disease (IBD) [31]. Currently, cyclic RGD (Arg-Gly-Asp) peptides have been shown to exhibit high affinity for integrin αvβ3 under inflammatory conditions [32], and this integrin is highly expressed on the surface of activated macrophages and endothelial cells [33], thereby playing a key role in macrophage-dependent inflammation [34,35]. These engineered nanovesicles can be functionalized to actively target inflamed lung tissues and specifically recognize activated macrophages and damaged endothelial cells during inflammation. Moreover, exogenously supplemented cholesterol can enhance vesicle stability and achieve improved drug loading, conferring reliability for their use as nanomedicine delivery platforms with rapid delivery and stable targeting [36].

In this study, we developed a multifunctional drug delivery platform, GRb1@LEVs-cRGD, for enhanced targeted therapy of ALI and ARDS (Scheme 1). In this system, GRb1 is fused with LEVs to generate hybrid nanovesicles that exert coordinated anti-inflammatory and anti-apoptotic activities. To achieve precise targeting, cRGD peptides were functionalized on the vesicle surface, facilitating delivery to immune cells, epithelial cells, and endothelial cells, thereby improving therapeutic efficacy in murine models of ALI/ARDS. Following intravenous administration, GRb1@LEVs-cRGD preferentially accumulates in inflamed lung tissues, exerting combined benefits of GRb1 and LEVs. Specifically, the platform maintains macrophage polarization homeostasis by inhibiting NF-κB activation, while simultaneously protecting immune and endothelial cells from apoptosis and dysfunction, preventing immune exhaustion and endothelial injury—two critical factors in ALI/ARDS pathogenesis. Moreover, cholesterol in Chol-PEG2000-cRGD integrates into the vesicle membrane, stabilizing the pH gradient and enhancing drug loading via a remote active loading strategy. Consequently, the loading efficiencies of tigecycline (TIG) and vancomycin (Vanc) increased six-fold following cholesterol supplementation. In murine models of bacterial pneumonia induced by carbapenem-resistant Klebsiella pneumoniae (CRKP) and methicillin-resistant Staphylococcus aureus (MRSA), antibiotic-loaded GRb1@LEVs-cRGD effectively reduced bacterial burden, mitigated inflammatory cytokine storms, and alleviated lung injury, demonstrating its robust drug-loading and therapeutic potential. Overall, GRb1@LEVs-cRGD combines the intrinsic anti-inflammatory properties of LEVs with the cytoprotective and anti-apoptotic effects of GRb1, restoring immune balance, preserving epithelial-endothelial integrity, and significantly improving outcomes in severe ALI/ARDS, representing a promising strategy for clinical intervention.

Scheme 1.

Design strategy and the fabrication process of GRb1@LEVS-cRGD and its therapeutic effects on ALI. (a) The hybrid vesicle was constructed through fusion of lemon derived vesicles (LEVs) and nanosized ginsenoside Rb1 (GRb1). Subsequently, GRb1@LEVS-cRGD were fabricated by modifying Chol-PEG₂₀₀₀-cRGD onto the surface of GRb1@LEVs using a membrane insertion technique. Antibiotics TIG and Vanc was sequentially encapsulated via remote loading to produce TIG/GRb1@LEVS-cRGD and Vanc/GRb1@LEVS-cRGD, respectively. (b) In vivo, GRb1@LEVS-cRGD markedly attenuated the production of inflammatory mediators, suppressed pulmonary inflammation, and mitigated lung tissue damage. Antibiotic-loaded GRb1@LEVS-cRGD effectively reached the infection sites and exhibited both anti-inflammatory and synergistic bactericidal effects. (c) Cytoprotective Capacity of GRb1@LEVs-cRGD in inflammation.

2. Results and discussion

2.1. Synthesis and characterization of GRb1@LEVs-cRGD

Firstly, LEVs were isolated from fresh lemon juice via differential centrifugation following established protocols [19] [37] [38] [39]. Regarding the extraction efficiency, an average of 11.52 ± 2.66 mg of LEVs (based on protein weight) was obtained from 300 mL of fresh lemon juice. Transmission electron microscopy (TEM) images confirmed that the LEVs exhibited a characteristic cup-shaped, double-membrane vesicle structure (Fig. 1a). To further verify purity, Dynamic Light Scattering (DLS) measurements were conducted, revealing a hydrodynamic diameter of 147.80 ± 2.28 nm (Fig. 1e) and a low polydispersity index (PDI) of 0.15 ± 0.003 (n = 10). This low PDI value demonstrates superior batch uniformity and the absence of significant impurities. Additionally, Nanoparticle Tracking Analysis (NTA) showed a particle concentration of 8.80 × 10¹² particles mL⁻¹, with consistent yields across independent batches (ranging from 6.41 × 10¹² to 10.65 × 10¹² particles mL⁻¹) (Fig. S1). This high yield and stability provide a solid foundation for subsequent applications.

Fig. 1.

Synthesis and characterization of GRb1@LEVS-cRGD. (a) Transmission electron microscopy (TEM) images of lemon‐derived EVs (LEVs), GRb1, GRb1@LEVs, and GRb1@LEVs-cRGD. (b) Fluorescence emission spectra of GRb1@LEVs. LEVs was doped with DiD and DiI, and then mixed with increasing amount of GRb1. (c) CLSM image of GRb1@LEVs prepared from DiO-labeled GRb1 (green) and DiD-labeled LEVs (red). (d) CLSM image of GRb1@LEVs-cRGD prepared from Chol-PEG₂₀₀₀-cRGD-FITC (green) and DiD-LEVs (red). (e) Hydrodynamic size distribution of nanoparticles determined by DLS. f) Zeta potentials of LEVs, GRb1, GRb1@LEVs, and GRb1@LEVs-cRGD (n = 3). (g-h) The stability on size (g) and zeta potential (h) of GRb1@LEVs-cRGD in PBS or PBS containing 10% FBS medium for 5 weeks (n = 3). (i) Schematic illustration of antibiotics remote loading into vesicles. (j) TIG loading yield at different cholesterol inputs (n = 3). (k) Loading yield at different TIG input (n = 3).

Previous studies have reported that lemons are rich in various anti-inflammatory components that can inhibit the production of inflammatory factors and mitigate inflammatory responses [40]. To identify the active constituents in LEVs and verify their compositional stability, we performed non-targeted metabolomic sequencing. The analysis revealed 154 consistently expressed metabolites across all batches with highly reproducible expression profiles, indicating excellent stability in the metabolic composition of LEVs (Fig. S2a). These metabolites primarily belonged to organic acids and derivatives (28.3%), phenylpropanoids and polyketides (23.3%), and organoheterocyclic compounds (13.8%) (Fig. S2b). Furthermore, a series of flavonoids with known potent anti-inflammatory activity were identified, such as eriocitrin, hesperidin, neoeriocitrin, and diosmetin-7-O-neohesperidoside (Fig. S2c), providing a molecular basis for the anti-inflammatory effects of LEVs.

Subsequently, to obtain GRb1@LEVs-cRGD with enhanced targeting efficacy, we first fused the pre-synthesized nano-sized GRb1 with LEVs via sonication and incubation, yielding GRb1@LEVs. To enhance inflammation targeting and minimize the off-target effects, cRGD peptides, which specifically bind integrin αvβ3, were conjugated to GRb1@LEVs using a membrane insertion technique, mediated by Chol-PEG₂₀₀₀-cRGD. Transmission electron microscopy (TEM) images revealed that LEVs exhibit a typical cup-like shape of EVs, GRb1 presents a relatively regular spherical structure, while GRb1@LEVs and GRb1@LEVs-cRGD display a typical vesicular structure with a lipid bilayer. (Fig. 1a).

To verify the successful fusion process of GRb1@LEVs, LEVs were labeled with a pair of Förster resonance energy transfer (FRET) dyes, DiD and DiI, followed by the addition of unlabeled GRb1 at various weight ratios for ultrasonic treatment (GRb1/LEVs = 0:1, 1:1, 4:1, and 10:1). As the GRb1 concentration increased gradually, Fig. 1b revealed an obvious fluorescence recovery at 565 nm and a corresponding decrease at 670 nm. This indicates that the interaction between FRET pairs in the original LEVs is weakened due to the intercalation of GRb1. To further validate the successful integration of the two materials, LEVs were labeled with the fluorescent dye DiO (green), while GRb1 was labeled with DiD (red). The significant colocalization of fluorescent signals between DiO and DiD confirm the successful fusion of LEVs and GRb1 (Fig. 1c). To achieve enhanced targeting of this nanomedicine delivery platform to inflamed lung tissues, we performed cRGD modification. Surface modification with cRGD was confirmed by colocalization of Chol-PEG₂₀₀₀-cRGD-FITC and DiD-GRb1@LEVs signals via CLSM (Fig. 1d). Dynamic light scattering (DLS) analysis revealed that the hydrodynamic diameters of LEVs and GRb1 was 147.80 ± 2.28 nm and 15.46 ± 1.31 nm, which increased to 168.20 ± 2.58 nm and 178.10 ± 1.09 nm for GRb1@LEVs and GRb1@LEVs-cRGD, respectively, which is consistent with the increased size observed by TEM (Fig. 1e). Zeta potential measurements showed values of −25.23 mV for LEVs, −5.88 mV for GRb1, -29.52 mV for GRb1@LEVs, and −15.59 mV for GRb1@LEVs-cRGD, reflecting surface charge alterations following modification (Fig. 1f). These findings ultimately confirm the effective conjugation of cRGD peptides with LEVs hybridized with GRb1, thereby forming GRb1@LEVs-cRGD. Stability assays demonstrated that GRb1@LEVs-cRGD remained stable in PBS and 10% FBS for up to 5 weeks (Fig. 1g and h).

Subsequently, we explored the potential of GRb1@LEVs-cRGD as a bioinspired drug delivery platform. Briefly, Chol-PEG₂₀₀₀-cRGD was integrated into the vesicle membrane via hydrophobic interactions. Exogenously supplemented cholesterol was utilized to establish and maintain a stable pH gradient across the vesicle membrane. Mechanistically, cholesterol inserts into the lipid bilayer, modulating membrane fluidity and inducing tighter lipid packing. This reduces membrane permeability to protons, thereby preventing the dissipation of the transmembrane pH gradient which acts as the driving force for active drug loading. This gradient drove the passive diffusion of drug molecules across the phospholipid bilayer into the internal aqueous lumen of the vesicles. Subsequently, ionization of the drug molecules within the lumen enabled their effective entrapment and retention inside the vesicles, thereby achieving remote drug loading (Fig. 1i). In the experiment, we observed that the addition of 5 wt% cholesterol achieved the maximum drug-loading capacity, reaching 14.28 ± 1.79% of the vesicle protein mass (Fig. 1j). Furthermore, we investigated how the amount of TIG added affects the drug-loading efficiency. The results showed that the loading yield peaked when the TIG addition amount was 2.0 mg mL⁻¹ (Fig. 1k). The release profile of TIG/GRb1@LEVs-cRGD over 48 h showed an initial burst release of 13.82% within the first 2 h, followed by sustained release up to 24 h, with a cumulative release exceeding 77.37% (Fig. S3a). The cumulative release data closely fitted the Higuchi diffusion model (R² = 0.955), indicating that the release process was predominantly controlled by a diffusion mechanism. Furthermore, stability tests demonstrated that TIG/GRb1@LEVs-cRGD remained stable in both PBS and 10% FBS for up to 4 weeks (Fig. S3b). After storage at 4 °C for 4 weeks and undergoing twice freeze-thaw cycles, the system retained over 85% of the drug, demonstrating good formulation stability (Fig. S3c).

Accordingly, we also explored the effectiveness of active loading for another antibiotic, Vanc. Similarly, when 5 wt% cholesterol was added, the drug-loading amount of Vanc reached a maximum of 13.83 ± 0.66%, and the maximum loading yield was achieved at an addition amount of 2.0 mg mL⁻¹ (Fig. S4). Corresponding stability and release experiments indicated that Vanc/GRb1@LEVs-cRGD also possesses the potential for sustained drug release, along with good formulation stability (Fig. S5).

These results demonstrate that we successfully constructed the hybrid nanovesicle carrier GRb1@LEVs-cRGD, which exhibits excellent stability under various conditions and enhanced capacity for antibiotic loading, laying a foundation for its application in pneumonia treatment.

2.2. Good biocompatibility of GRb1@LEVs-cRGD in vitro and in vivo

To explore the biomedical potential of GRb1@LEVs-cRGD, we first investigated its biocompatibility in vitro. Specifically, RAW264.7, MLE-12, and HUVEC cells were treated with varying concentrations of GRb1, LEVs, GRb1@LEVs, and GRb1@LEVs-cRGD for 24 h. Subsequently, CCK-8 assays were performed to assess cytotoxicity. CCK-8 assay results (Fig. 2a–c) indicated that GRb1, LEVs, GRb1@LEVs, and GRb1@LEVs-cRGD exhibited negligible cytotoxicity toward all cell lines within the concentration range of 10–150 μg mL⁻¹ (where LEVs were quantified by protein concentration, and the remaining nanoparticles were expressed as GRb1 concentration). Based on the results of the CCK-8 assay, we further selected the aforementioned materials at a concentration of 100 μg mL⁻¹ for live/dead staining to verify their biosafety. The results showed that nearly 100% of cells remained viable (Calcein AM, green) with few dead cells (Propidium Iodide, red) across various cell lines (RAW264.7, MLE-12, and HUVEC cells) (Fig. 2d). In addition, Phalloidin/DAPI staining was used to investigate the morphology of co-cultured cells. The staining results indicated that no significant disruption of the cytoskeleton (Phalloidin, green) was observed in HUVECs at 12 h and 48 h after treatment with the aforementioned nanomaterials (Fig. 2e). In summary, when incubated with GRb1, LEVs, GRb1@LEVs, and GRb1@LEVs-cRGD, the viability and morphology of RAW264.7, MLE-12, and HUVEC cells remained healthy, consistent with those of the control group.

Fig. 2.

Good biocompatibility of GRb1@LEVs-cRGD in vitro and in vivo. (a-c) CCK-8 assays in RAW264.7, MLE-12 and HUVEC cells, respectively (n = 3). (d) Live/dead cell staining in RAW264.7, MLE-12 and HUVEC cells. Green, Calcein AM staining for live cells. Red, propidium iodide staining for dead cells (n = 3). (e) Cell skeleton staining in HUVECs. Green, FITC phalloidine. Blue, DAPI. (f) Blood routine and biochemical indexes (n = 3). (g) Representative TUNEL staining of lung tissues from GRb1@LEVs-cRGD and control groups (n = 3). (h) Histopathological examination by H&E staining of major organs (n = 3).

Given the promising biocompatibility observed in vitro, we further administered GRb1@LEVs-cRGD to mice via tail vein injection for seven consecutive days to systematically evaluate its in vivo safety. Hematological analysis revealed no significant differences between the GRb1@LEVs-cRGD group and the PBS group in terms of white blood cell, lymphocyte, monocyte, granulocyte, red blood cell, and platelet levels. Biochemical function assays showed comparable activities of liver function markers alanine transaminase (ALT) and aspartate transaminase (AST), along with essentially equivalent concentrations of renal function parameters blood urea nitrogen (BUN) and creatinine (CREA). These results indicate that GRb1@LEVs-cRGD exhibits no hepatotoxicity or nephrotoxicity. (Fig. 2f). Furthermore, TUNEL staining of lung tissues detected no significant increase in cell apoptosis (Fig. 2g), and H&E staining of major organs (heart, liver, spleen, lung, and kidney) revealed no evidence of histopathological damage or inflammation induced by GRb1@LEVs-cRGD administration (Fig. 2h). In addition, to rigorously assess the potential immunogenicity of GRb1@LEVs-cRGD, we measured the levels of complement activation products and various cytokines in mouse serum following consecutive intravenous administrations (daily for 7 days). The results showed that the serum C3a level in the GRb1@LEVs-cRGD group was comparable to that in the PBS control group, indicating no complement activation (Fig. S6a). Furthermore, the levels of key inflammatory cytokines (IL-6, TNF-α, IFN-α, and IL-10) showed no statistically significant difference compared to the control group, confirming that the treatment did not induce a systemic cytokine storm (Fig. S6b). To further evaluate hemocompatibility for intravenous use, a hemolysis assay was conducted. The results demonstrated negligible hemolysis (<5%) even at high concentrations, indicating excellent blood compatibility (Fig. S6c).

In conclusion, these findings highlight the favorable biosafety profile of GRb1@LEVs-cRGD. Leveraging these advantages and excellent biosafety, the biomimetic drug delivery system GRb1@LEVs-cRGD constructed by us is expected to become a preferred option for future ALI/ARDS treatment.

2.3. In vitro and in vivo targeted delivery of GRb1@LEVs-cRGD

To evaluate the inflammatory lung-targeting capability of GRb1@LEVs-cRGD, DiD-labeled vesicles (DiD-GRb1@LEVs and DiD-GRb1@LEVs-cRGD) were co-incubated with RAW264.7, HUVEC and MLE-12 cells to assess cellular uptake in activated macrophages, endothelial cells and epithelial cells. CLSM analysis revealed that no difference in uptake was observed between GRb1@LEVs-cRGD and unmodified GRb1@LEVs in resting RAW264.7 cells. However, under inflammatory stimulation, activated RAW264.7 cells exhibited a significant increase in the uptake of GRb1@LEVs-cRGD, showing the highest intracellular DiD fluorescence, which was significantly greater than that of GRb1@LEVs without cRGD modification. Moreover, there was a marked difference in the internalization of GRb1@LEVs-cRGD between physiological and inflammatory states: compared with the physiological state, the internalization of GRb1@LEVs-cRGD by RAW264.7 cells was significantly enhanced under inflammatory conditions (Fig. 3a, Fig. S7). Subsequently, a similar experimental design was employed to evaluate the targeted uptake capacity of GRb1@LEVs-cRGD in HUVECs and MLE-12 cells. It was found that in the GRb1@LEVs-cRGD-treated group, the fluorescence in inflamed HUVECs increased significantly compared with that in resting HUVECs. In contrast, no significant difference in fluorescence intensity was observed between cells treated with GRb1@LEVs. Meanwhile, under inflammatory stimulation, consistent with the fluorescence images of macrophages, the internalization capacity of GRb1@LEVs-cRGD was significantly stronger than that of GRb1@LEVs, and this difference was not observed in resting HUVECs (Fig. 3b, Fig. S7). Similarly, this targeting effect was also observed in the MLE-12 pulmonary epithelial cell model, where the endocytic capacity of GRb1@LEVs-cRGD was significantly enhanced under inflammatory stimulation (Fig. S8a). Quantitative analysis further confirmed this trend, demonstrating that under inflammatory and physiological conditions, the fluorescent area corresponding to GRb1@LEVs-cRGD uptake in RAW264.7, HUVEC and MLE-12 cells increased by 12.06%, 17.93% and 13.30%, respectively (Fig. 3c and d, Fig. S8b). This confirms the capability of the platform to target the inflamed epithelium. To further consolidate the specific interaction between the cRGD motif and integrin αvβ3, we performed competitive inhibition assays by pre-blocking the receptors with an excess of free cRGD peptide. As expected, the cellular uptake of DiD-labeled GRb1@LEVs-cRGD in RAW 264.7, HUVEC, and MLE-12 cells under inflammatory stimulation was significantly reduced after pre-blocking (Fig. S9). This significant inhibition indicates the critical role of integrin αvβ3 in mediating the targeted internalization of the vesicles. These findings indicate that cRGD peptide-modified GRb1@LEVs-cRGD is more efficiently internalized by LPS-stimulated macrophages and endothelial cells, which facilitates the further in vivo application of the material [33].

Fig. 3.

In vitro and in vivo targeted delivery of GRb1@LEVs-cRGD (a-b) confocal microscopy images of the uptake of nanoparticles by RAW 264.7 (a) and HUVEC cells (b) under inflammation and physiological conditions. (c-d) The corresponding quantitative analysis in RAW 264.7 (c) and HUVEC cells (d), respectively (n = 5). (e-f) IVIS images showing the fluorescent distribution in the heart, liver, spleen, lung, and kidney at 2 h after injection of Cy5.5-labeled vesicles (e), along with the corresponding quantification of mean fluorescent intensity (f) (n = 3). (g) Corresponding fluorescence intensity quantification of lung at different times (n = 3). (h) Uptake distribution of GRb1@LEVs-cRGD in various lung cell types (n = 3).

Inspired by in vitro results, we used an in vivo imaging system (IVIS) to evaluate the biodistribution of Cy5.5-GRb1@LEVs or Cy5.5-GRb1@LEVs-cRGD in ALI mice. 2 h after tail vein injection of Cy5.5-labeled vesicles, a prominent fluorescent signal was observed in the lungs of ALI mice, which was distinctly different from the fluorescent signal in healthy mice (Fig. 3e). Notably, Cy5.5-GRb1@LEVs-cRGD exhibited significantly enhanced accumulation in inflamed lungs and reduced hepatic accumulation compared to unmodified GRb1@LEVs (Fig. 3e). Quantitative analysis of mean fluorescence further confirmed that the accumulation of GRb1@LEVs-cRGD in inflamed lung tissues was significantly increased compared with that in healthy lungs (Fig. 3f). Furthermore, this advantage persisted at 12 h post-injection, indicating that targeted vesicles can preferentially home to inflamed sites while evading non-specific immune clearance (Fig. 3g, Fig. S10). Furthermore, we systematically investigated the uptake of GRb1@LEVs-cRGD by various pulmonary cell types within the injured lung tissue. After digesting the lung tissue, cells were stained with antibodies against CD326 (epithelial cells), CD31 (endothelial cells), CD45 and F4/80 (macrophages), respectively. As shown in Fig. 3h and Fig. S11, in the Cy5.5-labeled GRb1@LEVs-cRGD group, the percentage of Cy5.5-positive cells reached 60.77% in macrophages, 27.03% in epithelial cells, and 21.46% in endothelial cells. These results demonstrate that GRb1@LEVs-cRGD can efficiently accumulate in lung tissue and be effectively taken up by various pulmonary cells, including macrophages, epithelial cells, and endothelial cells. Notwithstanding, substantial future investigations are warranted to further refine the targeting efficiency of these nanocarriers, with the aim of augmenting their therapeutic efficacy.

In conclusion, these findings verify that GRb1@LEVs-cRGD possesses robust in vitro and in vivo targeting capabilities. Surface-modified cRGD enhances the targeting capability of GRb1@LEVs-cRGD, enabling it to selectively accumulate at inflammatory sites. This selective accumulation strengthens its anti-inflammatory properties and cytoprotective capacity, thereby enhancing the therapeutic potential of GRb1@LEVs-cRGD for inflammatory lung diseases such as ALI/ARDS.

2.4. The anti-inflammatory capacity of GRb1@LEVs-cRGD in vitro

Macrophage polarization, as a key factor in the progression of ALI, plays a critical role in regulating inflammatory status, promoting lung injury recovery, and facilitating pulmonary barrier repair. To explore the potential of GRb1@LEVs-cRGD in influencing macrophage polarization, we evaluated the expression of M1 and M2 macrophage markers in RAW264.7 cells using flow cytometry. As illustrated in Fig. 4a and Fig. S12, an increased expression of CD86 (a marker for M1 macrophages) was observed following LPS stimulation, while the proportion of CD86⁺ cells decreased after drug treatment. Specifically, the proportions of CD86⁺ cells in the GRb1 and LEVs groups were reduced to 16.7% and 18.1%, respectively, whereas the optimized GRb1@LEVs-cRGD further lowered this proportion to 7.07%, which was significantly lower than the 27.0% observed in the model group. In contrast, the expression of CD206 (a marker for M2 macrophages) exhibited an opposite pattern: the GRb1@LEVs-cRGD group showed the highest M2 macrophage content (19.1%), which was significantly higher than that in the model group (1.18%). The single components GRb1 and LEVs exerted only a moderate effect on M1 to M2 macrophage polarization. Thus, compared to other groups, GRb1@LEVs-cRGD exhibited the most effective shift in macrophage phenotype from M1 to M2. To corroborate these findings at the transcriptional level, the mRNA expression of macrophage polarization markers was assessed using qRT-PCR. Consistent with the flow cytometry data, Fig. S13 shows that the mRNA expression of the M1 marker (CD86) and pro-inflammatory cytokines (TNF-α, IL-6) increased significantly after LPS stimulation but was markedly suppressed following GRb1@LEVs-cRGD treatment. Conversely, the transcriptional levels of M2-associated genes—CD206, CD163, and Arg-1—were downregulated in the model group, while GRb1@LEVs-cRGD intervention significantly restored their expression.

Fig. 4.

The anti-inflammatory effect of the nanoparticles in vitro (a) The proportion of CD86 M1 and CD206 M2 in RAW264.7 cells by flow cytometry. (b-c) Level of inflammatory cytokines IL-6, TNF-α and IL-10 in LPS-stimulated RAW 264.7 cells (b) and MH-S cells (c) after different treatments (n = 5). (d-e) NF-κB p65 nuclear translocation observed by CLSM (n = 5). ①control group; ② PBS treated group; ③ LEVs treated group; ④ GRb1 treated group; ⑤ GRb1@LEVs treated group; ⑥ GRb1@LEVs-cRGD treated group. (f) The protein expressions of key members in the NF-κB pathway by Western blot, including the phosphorylated (p-p65) and basal NF-κB p65, p-IκBα, and IκBα (n = 3). (g) Heat map showing the hierarchical clustering results of the DEGs detected between LPS group and LPS treated with GRb1@LEVs-cRGD group. DEGs were identified based on a fold change greater than 1.5 and an adjusted P-value less than 0.05. (g) Gene Ontology (GO) term enrichment analysis was performed, and the top 30 significantly enriched GO terms were selected based on an FDR <0.05. (h) Top 20 enriched pathways identified using Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway analysis of DEGs in cells triggered with LPS. (i-j) Gene Set Enrichment Analysis (GSEA) images. (l) Schematic illustration of the mechanism in the reprogramming of RAW 264.7 cells by GRb1@LEVs-cRGD.

Correspondingly, this shift would further extend to the secretion of pro-inflammatory/anti-inflammatory cytokines downstream of the pathway. Therefore, we evaluated the levels of pro-inflammatory cytokines (TNF-α and IL-6) and the anti-inflammatory cytokine IL-10 using enzyme-linked immunosorbent assay (ELISA). Compared to individual components, GRb1@LEVs-cRGD significantly reduced the secretion of proinflammatory cytokines IL-6 and TNF-α in both RAW264.7 (Fig. 4b) and MH-S cells (Fig. 4c), while upregulating anti-inflammatory cytokine IL-10 levels, indicating its regulatory effect on inflammatory responses. GRb1@LEVs-cRGD outperformed GRb1@LEVs, potentially attributed to RGD peptide-mediated enhancement of nanoparticle internalization by immune cells, which suppressed proinflammatory macrophage activation and mitigated cytokine storms. In addition, given that oxidative stress also plays a critical role in macrophage polarization during inflammation-related diseases, we further detected intracellular reactive oxygen species (ROS) levels in RAW264.7 cells following GRb1@LEVs-cRGD treatment. Compared to the individual components GRb1 and LEVs, the fused hybrid vesicles GRb1@LEVs exhibited additive antioxidant activity. After modification with cRGD peptides, the intracellular ROS-scavenging capacity of GRb1@LEVs-cRGD is significantly strengthened, reducing ROS levels to baseline. These findings were confirmed by corresponding quantitative analyses using a fluorescence microplate reader (Fig. S14).

The aforementioned data prompted us to explore the molecular mechanism underlying the anti-inflammatory effects of GRb1@LEVs-cRGD. First, the NF-κB pathway plays a central role in activating inflammatory responses, and the nuclear translocation of p65 is a critical step in this pathway. Thus, IF staining was performed to observe the distribution of p65 in the nuclei of RAW264.7 cells exposed to LPS. The results showed that LPS exposure triggered the translocation of p65 from the cytoplasm to the nucleus, and GRb1@LEVs-cRGD most effectively inhibited this process compared to GRb1 or LEVs used alone (Fig. 4d and e). To further elucidate the specific molecular events upstream of nuclear translocation, Western blot analysis was performed to detect the phosphorylation status of key signaling proteins. As anticipated, the phosphorylation levels of IκBα (Ser32/36) and p65 (Ser536) were significantly elevated in the LPS model group compared with the control. In contrast, GRb1@LEVs-cRGD treatment—particularly when compared with GRb1@LEVs—effectively reversed the hyperphosphorylation of these proteins. This suggests that the nanovesicles inhibit the degradation of IκBα and the subsequent phosphorylation/activation of p65, thereby blocking the NF-κB signaling cascade (Fig. 4f, Fig. S15).

Subsequently, transcriptomic analysis was performed to further evaluate changes in mRNA levels in LPS-treated RAW264.7 cells before and after GRb1@LEVs-cRGD administration. Differentially expressed genes (DEGs) were identified using rigorous screening thresholds of |log2(FC)| > 1 and p-value <0.05. Cluster analysis revealed that, compared with the untreated group, the GRb1@LEVs-cRGD treatment group exhibited 395 upregulated genes and 816 downregulated genes (Fig. 4g). GO enrichment analysis demonstrated reductions in genes associated with inflammatory responses (GO:0006954), innate immune responses (GO:0045087), immune system processes (GO:0002376), and cytokine activity (GO:0005125) (Fig. 4h). These biological processes are predominantly linked to inflammation, thereby confirming the anti-inflammatory effect of the combinatory therapy at the molecular level (red rectangle). Furthermore, the top 20 enriched pathways identified via Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway analysis were associated with multiple inflammatory pathways, including TNF (gene ID: mmu04668), nuclear factor kappa-B (NF-κB, gene ID: mmu04064), cytokine-cytokine receptor interaction (mmu04060), interleukin-17 (IL-17, gene ID: mmu04657), Toll-like receptor (gene ID: mmu04620), and NOD-like receptor (gene ID: mmu04621) signaling (Fig. 4i). Additionally, GSEA and KEGG enrichment analyses showed downregulation of TNF-α (mmu04668) and NF-κB (mmu04064) signaling pathways (Fig. 4j and k), consistent with the aforementioned findings.

In summary, the combined data demonstrate that GRb1@LEVs-cRGD inhibits the phosphorylation and degradation of IκBα, blocks the nuclear translocation of NF-κB p65, thereby downregulating the NF-κB-mediated pro-inflammatory signaling pathway, and ultimately suppresses the release of downstream inflammatory cytokines such as TNF-α, IL-1β, and IL-6 (Fig. 4l). These mechanisms suggest that this delivery system holds potential therapeutic value for inflammatory diseases including ALI.

2.5. Cytoprotective capacity of GRb1@LEVs-cRGD in inflammation

In addition to imbalanced macrophage polarization and uncontrolled inflammation, ALI is often triggered or exacerbated by disruption of the air-blood barrier [41] [42]. Following the demonstration of GRb1@LEVs-cRGD's remarkable anti-inflammatory capacity, we next investigated its cytoprotective effects under inflammatory conditions. Here, MLE-12 and HUVEC cells were used to evaluate the cytoprotective effects of GRb1@LEVs-cRGD on endothelial and epithelial cells. First, we induced cellular inflammatory injury using LPS, which significantly impaired cell viability. Among the various treatment groups, GRb1@LEVs-cRGD exhibited enhanced cytoprotective effects compared to GRb1 or LEVs used alone, significantly increasing the viability of MLE-12 cells from 49.3% to 82.7% (Fig. 5a). Similarly, the protective effect of GRb1@LEVs-cRGD was validated in HUVECs (Fig. 5b). As shown in Fig. 5c and d, GRb1@LEVs-cRGD reduced the content of Caspase-3 after LPS stimulation, leading us to hypothesize that it may exert anti-apoptotic effects by downregulating Caspase-3. Inflammation-induced apoptosis is a key factor leading to multi-organ damage and immunosuppression during ALI/ARDS. Subsequently, to further verify the anti-apoptotic capacity of GRb1@LEVs-cRGD, we investigated LPS-induced cell apoptosis and necrosis using flow cytometry. The results demonstrated that the introduction of GRb1@LEVs-cRGD significantly reduced the proportion of LPS-induced apoptotic and necrotic cells, providing evidence for its cytoprotective properties (Fig. 5e–g). Mitochondria play a pivotal role in regulating cell death, and a reduction in mitochondrial membrane potential (MMP) is widely recognized as a hallmark of apoptotic cell death. 5,5′,6,6′-Tetrachloro-1,1′,3,3′-tetraethylbenzimidazolylcarbocyanine iodide (JC-1) accumulates in the mitochondrial matrix of cells with high MMP, forming J-aggregates that emit intense red fluorescence. Following LPS treatment, red fluorescence diminished while green fluorescence increased, indicating reduced MMP. In injured MLE-12 and HUVEC cells, GRb1 and LEVs partially prevented the loss of mitochondrial membrane potential, whereas the therapeutic effect of GRb1@LEVs-cRGD was the most prominent: it maximally attenuated the increase in green fluorescence, prevented MMP depolarization, and thus demonstrated superior protective effects against cell necrosis and apoptosis (Fig. 5h and i, Fig. S16).

Fig. 5.

Cytoprotective Capacity of GRb1@LEVs-cRGD in inflammation. (a-b) Cell viability of MLE-12 (a) and HUVEC cells (b) under different treatment conditions (n = 5). (c-d) Caspase 3 activity of MLE-12 (c) and HUVEC (d) cells under different treatment conditions (n = 5). (e-g) Flow cytometry analysis using Annexin V/PI staining (e) and corresponding quantitative comparison of MLE-12 cells (f) and HUVEC cells (g) exposed to different treatments (n = 3). (h-i) Fluorescence images of JC-1-stained MLE-12 cells after different treatments (n = 5). (j) The expression pattern of ZO-1 protein by immunofluorescent staining in HUVECs (n = 5). (k-l) Fluorescence images (k) of TUNEL staining RAW 264.7 cells after different treatments (n = 5) and corresponding quantification (l) (n = 5).

Zonula occludens-1 (ZO-1) is a key peripheral membrane protein that forms tight junctions and maintains the integrity of vascular endothelium. ZO-1 localizes at the site of adhesion junctions during early formation and is crucial for preserving the air-blood barrier during the progression of ALI. IF images revealed that LPS stimulation resulted in a significant reduction and irregular arrangement of ZO-1 protein, along with disrupted endothelial permeability (Fig. 5j & S17). However, the addition of GRb1@LEVs-cRGD effectively prevented ZO-1 dissociation and endothelial barrier disruption, and its protective capacity was even stronger than that of GRb1 alone. Beyond epithelial and endothelial cells, GRb1@LEVs-cRGD also prevented LPS-triggered cell death in RAW264.7 cells. TUNEL staining is commonly used to assess apoptosis by detecting DNA fragmentation, a typical characteristic of apoptotic processes. In the LPS-treated model group, the average fluorescence intensity of TUNEL staining (green) was the highest. However, pretreatment of cells with GRb1, LEVs, GRb1@LEVs, or GRb1@LEVs-cRGD effectively attenuated the increase in green fluorescence, among which GRb1@LEVs-cRGD exhibited the strongest ability to reverse apoptosis (Fig. 5k and l). Subsequently, images obtained from cell viability/death staining of RAW264.7 cells using calcein AM/PI further validated the protective capacity of GRb1@LEVs-cRGD (Fig. S18). Moreover, consistent results were observed in the detection of mitochondrial membrane potential: GRb1@LEVs-cRGD most effectively prevented the loss of MMP, thereby demonstrating the strongest cytoprotective capacity that surpasses that of GRb1 or LEVs alone (Fig. S19).

In conclusion, these results demonstrate that GRb1@LEVs-cRGD effectively protects epithelial and endothelial cells from damage in vitro, maintains the stability of the alveolar-capillary barrier, and extends its therapeutic benefits to multiple cell types within the inflammatory microenvironment. This further highlights the excellent therapeutic potential of GRb1@LEVs-cRGD for ALI/ARDS.

2.6. GRb1@LEVs-cRGD alleviates LPS-induced lung injury

Building on the established biocompatibility and targeted delivery capability of GRb1@LEVs-cRGD, we further investigated its in vivo therapeutic efficacy using a mouse model of severe ALI (Fig. 6a). The ALI/ARDS mouse model was established by intratracheal administration of LPS (4 mg kg⁻¹). Subsequently, PBS, GRb1, LEVs, GRb1@LEVs, and GRb1@LEVs-cRGD were administered via tail vein injection at 1 h post-modeling, with both GRb1 and LEVs dosed at 8 mg kg⁻¹.

Fig. 6.

GRb1@LEVs-cRGD alleviated LPS-induced lung injury and inflammatory responses in the lung. (a) Schematic illustration of animal experimental design. (b-d) lung wet/dry ratio (b), levels of protein (c), total cell number (d) in BALF in LPS induced ALI mice treated with PBS, LEVs, GRb1, GRb1@LEVs, GRb1@LEVs-cRGD (n = 5). (e-h) Level of IL-6 and TNF-α in BALF (e) and serum (h) of mice after different treatments (n = 5). (i-j) Representative H&E images of lungs after different treatments (i) and the corresponding analysis of lung injury score (j) (n = 5). (k) Lung tissue analysis TUNEL staining in each group (n = 5). (l-n) The distribution of tight junction proteins, ZO-1 and VE-Cadherin, by immunofluorescent staining in lung tissues and corresponding quantification.

LPS stimulation triggers endothelial and epithelial barrier dysfunction, which, coupled with the upregulation of pro-inflammatory cytokines, leads to lung tissue damage and pulmonary edema. At 24 h after LPS stimulation, bronchoalveolar lavage fluid (BALF), blood, and lung tissues were collected for analysis. The wet/dry lung weight ratio, a key marker of pulmonary edema, showed that LPS injection induced significant pulmonary edema, while both GRb1@LEVs and GRb1@LEVs-cRGD treatments markedly alleviated pulmonary edema in mice (Fig. 6b). Owing to the superior inflammatory lung-targeting ability of GRb1@LEVs-cRGD, its inhibitory effect on pulmonary edema was superior to that of GRb1@LEVs. Furthermore, compared with free GRb1 and LEVs, GRb1@LEVs-cRGD also significantly reduced the total cell count and total protein concentration in BALF, indicating that the integrated delivery of GRb1 and LEVs enhanced the efficacy of both components. Through their synergistic action, LPS-induced lung barrier damage was effectively mitigated (Fig. 6c & d).

Notably, ALI is characterized by a cytokine-driven hyperinflammatory phase. Therefore, the levels of pro-inflammatory cytokines (IL-6 and TNF-α) in BALF and serum of each group were measured using ELISA to further confirm the anti-inflammatory capacity of GRb1@LEVs-cRGD. Compared with the PBS-treated group, administration of other therapeutic agents significantly reduced the concentrations of TNF-α and IL-6. Among them, the levels of proinflammatory cytokines IL-6 and TNF-α in BALF and serum of mice treated with GRb1@LEVs-cRGD were reduced to the lowest, even reaching baseline levels (Fig. 6e–h). These results indicate that GRb1@LEVs-cRGD, by integrating the synergistic effects of GRb1 and LEVs, inhibits the massive infiltration of inflammatory cells into alveoli and the release of inflammatory factors, thereby reversing LPS-induced lung tissue damage and pulmonary edema.

To investigate the impact of GRb1@LEVs-cRGD on the polarization homeostasis of M1/M2 macrophages in vivo, we further analyzed the expression of M1/M2 phenotypic markers and their related cytokines in lung tissue using immunofluorescence staining and qRT-PCR. IF staining results showed that in the LPS-induced ALI model, the expression of the pro-inflammatory M1 marker CD86 was significantly increased, while its expression decreased after GRb1@LEVs-cRGD treatment. Conversely, the anti-inflammatory M2 marker CD206 exhibited an opposite trend (Fig. S20). qRT-PCR results further demonstrated that, compared to the model group, GRb1@LEVs-cRGD treatment downregulated the transcriptional levels of the M1-related marker CD86 and its released inflammatory factors (TNF-α, IL-6), while upregulating the mRNA expression of M2-related markers CD206, CD163, and Arg-1 (Fig. S21). These findings indicate that GRb1@LEVs-cRGD can shift the pulmonary microenvironment from a pro-inflammatory to an anti-inflammatory state by correcting the imbalance in M1/M2 macrophage polarization, thereby suppressing the pulmonary inflammatory response and significantly alleviating lung injury.

Furthermore, histological analysis using H&E staining was performed to evaluate morphological changes in lung tissues. The results demonstrated that GRb1@LEVs-cRGD treatment effectively alleviated lung tissue inflammation. Notably, while GRb1, LEVs, and GRb1@LEVs all inhibited inflammatory cell infiltration, GRb1@LEVs-cRGD was deemed more effective than the other three groups in treating ALI/ARDS, as it significantly reduced pulmonary edema, alveolar wall thickening, and accumulation of foamy macrophages, with lung tissue structure highly resembling that of the normal control (Fig. 6i). Pathological scoring based on H&E staining further quantified the severity of LPS-induced acute lung injury, and the results showed that GRb1@LEVs-cRGD significantly ameliorated LPS-induced lung injury in ALI mice (Fig. 6j). Subsequently, we performed IF staining on lung samples to observe the expression of inflammatory cytokines. As shown in Figs. S22 and S23, mice in the control group exhibited the least infiltration of pro-inflammatory cytokines. In contrast, ALI led to a massive influx of inflammatory cells and excessive release of inflammatory cytokines, ultimately resulting in lung damage. Owing to the synergistic anti-inflammatory effects of GRb1 and LEVs, as well as the superior inflammatory lung targeting mediated by cRGD, the GRb1@LEVs-cRGD treatment group showed a significant reduction in inflammatory cytokines.

Furthermore, extensive histiocytic apoptosis in the lungs and damage to the pulmonary air-blood barrier are key determinants of the severity of ALI/ARDS [43]. Therefore, lung barrier integrity was evaluated via TUNEL staining of lung tissues and immunofluorescence assays for ZO-1 and VE-cadherin. As shown in Fig. 6k, ALI mice in the PBS-treated group exhibited a significant increase in apoptotic cells (red signals) in lung tissues, whereas GRb1@LEVs-cRGD treatment markedly reduced the number of apoptotic cells (red), highlighting the superior anti-apoptotic effect of GRb1@LEVs-cRGD. IF staining and quantitative analysis revealed that GRb1@LEVs-cRGD treatment rescued the downregulation of ZO-1 and VE-cadherin expression, indicating the restoration of lung barrier integrity (Fig. 6l–n).

In the long-term observation (7 days), GRb1@LEVs-cRGD demonstrated the potential to significantly improve the prognosis of ALI/ARDS. Survival analysis indicated that this treatment markedly enhanced the survival rate of mice compared to the model group. Arterial blood gas analysis further confirmed that GRb1@LEVs-cRGD restored oxygenation levels (PaO₂/FiO₂) (Fig. S24a). Furthermore, pulmonary function parameters (Penh, tidal volume, and EF50) measured by non-invasive whole-body plethysmography revealed that the treatment effectively alleviated LPS-induced lung barrier injury and promoted functional recovery (Fig. S24b–e). Histologically, H&E and Masson staining showed that GRb1@LEVs-cRGD mitigated lung tissue inflammation and preserved alveolar structure, contributing to an improved overall outcome (Fig. S24f–g).

Collectively, these findings demonstrate that GRb1@LEVs-cRGD exerts a positive effect in protecting mice against ALI/ARDS: through synergistic action, it reduces lung injury, improves lung function, effectively inhibits inflammatory storms in lung tissues, and maintains pulmonary epithelial-endothelial integrity. In vivo results further underscore the robust potential of GRb1@LEVs-cRGD in treating ALI.

2.7. Therapeutic efficacy of GRb1@LEVs-cRGD in Kp NDM-induced severe pneumonia

Currently, Carbapenem-resistant Klebsiella pneumoniae (CRKP) is recognized as one of the most severe nosocomial pathogens threatening public health [44] [45] [46]. Clinically, pulmonary bacterial infections often precipitate severe cytokine storms, posing life-threatening risks to patients [47]. Uncontrolled host inflammatory responses induced by bacteria can lead to ALI, which may rapidly progress to ARDS with a mortality rate of 30–50% [48]. Thus, the rapid elimination of pulmonary bacteria and restoration of host homeostasis are critical for developing therapies against lung infections. Consequently, we employed a pH-gradient strategy to encapsulate TIG within GRb1@LEVs-cRGD, a nanovesicle system that exhibits remarkable anti-inflammatory properties, to assess the therapeutic efficacy of TIG-loaded hybrid vesicles (TIG/GRb1@LEVs-cRGD) against CRKP (Fig. 7a).

Fig. 7.

TIG/GRb1@LEVs-cRGD alleviated Kp NDM-induced lung injury and inflammatory responses in the lung. (a) Experimental design of in vivo assessment using a Kp NDM-induced model. (b) Growth curves of KP NDM co-incubated with various preparations (n = 3). (c) Corresponding quantification of bacterial load in lung tissue homogenate (n = 5). (d-e) Lung wet/dry ratio (d) and levels of protein (e) in Kp NDM-induced ALI mice treated with PBS, TIG, TIG/GRb1@LEVs-cRGD and negative control (n = 5). (f-h) IL-6 (f) and TNF-α (g) IL-1β (h) of BALF in above groups (n = 5). (i-j) Representative H&E images of the lung after different treatments (i) and corresponding lung injury score analysis (j) (n = 5). (k-p) Immunofluorescence staining images of IL-6 (k) and corresponding quantitative analysis (n), TNF-α (l) and corresponding quantitative analysis (o), IL-1β(m) and corresponding quantitative analysis (p) (n = 5).

In this study, we first evaluated the in vitro antibacterial activity of TIG/GRb1@LEVs-cRGD against a CRKP strain of Kp NDM. Bacterial growth assays showed that TIG/GRb1@LEVs-cRGD exhibited antibacterial activity comparable to free TIG, with minimum inhibitory concentrations (MICs) similar to those of free TIG against Kp NDM (Fig. 7b, Fig. S25a). An acute severe bacterial pneumonia model was established in BALB/c mice through endotracheal intubation with clinical isolates Kp NDM. One hour after infection with Kp NDM bacterial suspension, mice were intravenously administered PBS, free TIG (15 mg kg⁻¹), or TIG/GRb1@LEVs-cRGD (15 mg kg⁻¹). Compared to the PBS group, the TIG/GRb1@LEVs-cRGD treatment group showed a ∼2.29 log CFU reduction in lung bacterial burden, and a ∼0.46 log CFU decrease relative to free TIG, indicating effective suppression of bacterial replication in lung tissues (Fig. 7c, Fig. S25b). Treatment with TIG/GRb1@LEVs-cRGD also significantly reduced lung wet/dry (W/D) ratios (Fig. 7d) and bronchoalveolar lavage fluid (BALF) protein levels (Fig. 7e), reflecting alleviated pulmonary edema. Additionally, levels of pro-inflammatory cytokines (e.g., IL-6, TNF-α, IL-1β) in BALF were downregulated compared to the PBS or free TIG groups (Fig. 7f–h). Histological analysis revealed that TIG/GRb1@LEVs-cRGD improved lung tissue by mitigating edema and reducing alveolar immune cell infiltration, yielding the lowest lung injury scores (Fig. 7i and j). IF staining of lung sections further confirmed reduced levels of inflammatory cytokines (IL-6, TNF-α, IL-1β), consistent with quantitative analyses (Fig. 7k–p). Additionally, to evaluate the long-term therapeutic efficacy in severe bacterial pneumonia, a 7-day longitudinal study was conducted in the CRKP-infected mouse model. The results demonstrated that TIG/GRb1@LEVs-cRGD significantly increased the survival rate (Fig. S26a) and improved both oxygenation index and pulmonary function parameters (Fig. S26b–e). H&E and Masson staining further confirmed that the treatment effectively resolved lung tissue inflammation and prevented structural deterioration, thereby enhancing the overall prognosis of severe CRKP pneumonia (Fig. S26f–g).

Collectively, TIG/GRb1@LEVs-cRGD demonstrates potent therapeutic potential in treating KP NDM-induced severe pneumonia in mice, effectively reducing bacterial load, inflammation, and lung injury.

2.8. Therapeutic efficacy of GRb1@LEVs-cRGD against MRSA-infected bacterial pneumonia in vivo

Associated with the increased use of antibiotics, the incidence of pulmonary infections caused by MRSA has sharply escalated, particularly in ventilator-associated hospital-acquired pneumonia, which is characterized by high morbidity and fatal infection rates [49] [50] [51] [52]. To explore additional therapeutic possibilities for MRSA, we incorporated the classic antimicrobial Vanc into GRb1@LEVs-cRGD vesicles, yielding Vanc/GRb1@LEVs-cRGD with a loading efficiency of 13.44% (Fig. 8a). In vitro antibacterial assays demonstrated that Vanc/GRb1@LEVs-cRGD exhibited antimicrobial performance comparable to free Vanc (Fig. 8b, Fig. S27a). Mice were then infected with MRSA and treated with PBS, free Vanc (8 mg kg⁻¹), or Vanc/GRb1@LEVs-cRGD (8 mg kg⁻¹). As anticipated, visual inspection of plates and direct bacterial counting revealed that Vanc/GRb1@LEVs-cRGD significantly inhibited bacterial infection in lung tissue homogenates, reducing the bacterial count by approximately 1.63 CFU (Fig. 8c, Fig. S27b).

Fig. 8.

Vanc/GRb1@LEVs-cRGD alleviated MRSA-induced lung injury and inflammatory responses in the lung. (a) Experimental design of in vivo assessment using a MRSA-induced model. (b) Growth curves of MRSA co-incubated with various preparations (n = 3). (c) Corresponding quantification of bacterial load in lung tissue homogenate (n = 5). (d-e) Lung wet/dry ratio (d) and levels of protein (e) in MRSA-induced ALI mice treated with PBS, Vanc, Vanc/GRb1@LEVs-cRGD and negative control (n = 5). (f-h) IL-6 (f) and TNF-α (g) IL-1β (h) of BALF in above groups (n = 5). (i-j) Representative H&E images of the lung after different treatments (i) and corresponding lung injury score analysis (j) (n = 5). (k-p) Immunofluorescence staining images of IL-6 (k) and corresponding quantitative analysis (n), TNF-α (l) and corresponding quantitative analysis (o), IL-1β (m) and corresponding quantitative analysis (p) (n = 5).

Correspondingly, treatment with Vanc/GRb1@LEVs-cRGD led to a notable decrease in the lung wet/dry ratio, indicating substantial alleviation of pulmonary edema (Fig. 8d). Additionally, protein levels and proinflammatory cytokiinnatenes (IL-6, TNF-α, and IL-1β) in BALF were significantly lower than those in the PBS group (Fig. 8e–h). H&E staining further confirmed the potent antibacterial and anti-inflammatory effects of Vanc/GRb1@LEVs-cRGD, showing reduced lung tissue damage, with lung injury scores consistent with these findings (Fig. 8i and j). IF staining of lung tissues further validated the reduction in inflammatory cytokines (Fig. 8k–p). Similarly, in the long-term MRSA pneumonia model, Vanc/GRb1@LEVs-cRGD demonstrated significant therapeutic potential. Survival analysis (n = 10) revealed a marked improvement in the 7-day survival rate (Fig. S28a). Consistent with the CRKP model, mice treated with Vanc/GRb1@LEVs-cRGD exhibited significant recovery in oxygenation and multiple pulmonary function indicators (Fig. S28b–e). Histopathological analysis (H&E and Masson staining) further corroborated these findings, showing substantially alleviated inflammatory infiltration and tissue damage (Fig. S28f–g).

Collectively, Vanc/GRb1@LEVs-cRGD demonstrated robust in vivo antibacterial and anti-inflammatory activities, further verifying the broad applicability and efficacy of this lung-targeted drug delivery platform.

Against the backdrop of the rapidly evolving field of nanomedicine for ALI/ARDS treatment, the design concept of our platform aligns well with the mainstream pursuit of efficient targeting and synergistic therapy. Recent research has focused on innovations in materials engineering, such as carrier-free nanodrugs [53] and biocompatible delivery strategies [54,55], to enhance biosafety and accumulation at inflammatory sites. For instance, Liao et al. [56] recently highlighted the potential of advanced nanocarriers in pulmonary inflammation modulation. In line with this trend, our work further emphasizes the dual advantages of “active carriers”: unlike inert synthetic carriers, the selected LEVs are rich in bioactive flavonoids, serving not only as vehicles but also contributing intrinsic anti-inflammatory properties. Furthermore, GRb1@LEVs-cRGD embodies a synergistic logic of “host immune modulation combined with pathogen clearance,” utilizing pH-gradient loading to target drug-resistant bacteria. This dual-mode strategy offers more comprehensive coverage than single-mode nanosystems, providing a significant foundation for developing next-generation nanotherapeutic platforms.

Finally, regarding the experimental design, it should be noted that this study employed only male mice to minimize the potential variability introduced by the estrous cycle and sex hormones on immune responses. However, given that sex differences can influence the pathophysiology of ALI/ARDS, future studies should include both sexes to comprehensively evaluate the efficacy and safety of GRb1@LEVs-cRGD and to explore potential sex-specific therapeutic outcomes.

3. Conclusion

In this study, we developed a plant-derived nanovesicle-based therapeutic platform, GRb1@LEVs-cRGD, for targeted intervention in ALI/ARDS. By integrating GRb1 with LEVs and functionalizing the surface with cyclic RGD peptides, this system combines the inherent anti-inflammatory activity and biocompatibility of plant-derived vesicles with targeted delivery capability. GRb1@LEVs-cRGD suppressed inflammatory cascades, rebalanced macrophage polarization, and preserved epithelial–endothelial barrier integrity in vitro, while exhibiting favorable safety profiles consistent with the edible origin of LEVs. In murine ALI/ARDS models, targeted delivery of GRb1@LEVs-cRGD reduced pulmonary edema, attenuated cytokine storms, mitigated histopathological damage, and promoted barrier repair. Transcriptomic analysis confirmed broad suppression of proinflammatory signaling pathways, supporting its potential applicability to other inflammatory lung diseases such as pneumonia. Cholesterol-mediated membrane engineering further enhanced drug-loading capacity, enabling efficient encapsulation of antibiotics and conferring combined antibacterial and anti-inflammatory activity against drug-resistant bacterial pneumonia. Collectively, these findings establish GRb1@LEVs-cRGD as a biocompatible, anti-inflammatory, and broadly applicable nanocarrier platform for precision treatment of severe inflammatory lung disorders.

4. Experimental section

4.1. Materials

Ginsenoside Rb1 and Tigecycline (TIG) were supplied by Shanghai Yuanye Bio-Technology Co., Ltd (China). Vancomycin-HCl (Vanc) was purchased from Aladdin Chemistry Co., Ltd (Shanghai, China). Chol-PEG₂₀₀₀-cRGD and Chol-PEG₂₀₀₀-cRGD-FITC were purchased from Shanghai ToYong Biotech Co., Ltd (China). Ammonium sulfate ((NH4)₂SO₄), dimethyl sulfoxide (DMSO, ≥99%) were from Shanghai Titan Scientific Co., Ltd (China). DiO (DiOC₁₈(3)), DiD (DiIC₁₈(5)), DiI (DiIC₁₈(3)), 2-(4-Amidinophenyl)-6-indolecarbamidine dihydrochloride (DAPI) were from Beyotime Biotechnology Co., Ltd (China). Lipopolysaccharides (LPS) was sourced from Sigma-Aldrich (USA). Fluorescein phalloidin and cyanine5.5 amine were procured from MedChemExpress (USA). 10% FBS-Premium was purchased from NEWZERUM Co., Ltd (China). Enhanced Cell Counting kit-8(CCK8), One Step TUNEL Apoptosis Assay Kit, enhanced mitochondrial membrane potential assay kit with JC-1, Caspase 3 Activity Assay Kit, Enhanced BCA Protein Assay Kit, BeyoDesalt™ G-25 Mini Desalting Column were from Beyotime Biotechnology Co., Ltd (Shanghai, China). Mouse precoated ELISA kits for TNF-α, IL-6, IL-1β were purchased from Beijing Dakewe Biotechnology Co., Ltd (China).NF-kB p65/RelA Rabbit mAb (A22331) and ZO-1 Rabbit mAb (A25306) were purchased from ABclonal Technology Co., Ltd (China).

4.2. Characterization

The morphology of nanoparticles was visualized by the field emission transmission electron microscopy (TEM, FEI Talos FX200X G2, Thermo Fisher, USA). Dynamic light scattering (DLS) and zeta potential were obtained by using a nanoparticle size and zeta potentiometer (Omni, Brookhaven Co., USA). Fluorescence spectroscopy (Edinburgh 3 Instruments. UK) was employed to verify the successful integration of GRb1 and LEVs. UV-vis absorption spectra (SHIMADZU, Japan) and microplate spectrophotometer (BioTek Epoch 2, Agilent, USA) were utilized to determine the loading concentrations of TIG and Vanc. Confocal images were obtained using a Leica STELLARIS 5 (China), the images were analyzed by Leica Application Suite X (LAS X). For fluorescence imaging of organs by using IVIS spectrum (PerkinElmer, USA).

4.3. Isolation and purification of EVs

EVs were isolated from lemon juice as previously described [19] [37] [38]. Briefly, the juice was differentially centrifuged (500×g for 10 min, 3000×g for 15 min, 5000×g for 30 min and 10,000×g for 1 h). The above supernatant was filtered at 0.8 and 0.45 μm pore filter and centrifuged at 16,500×g for 3 h. The supernatant was then ultra‐centrifuged at 100,000×g for 2 h with SW28 rotor, and the pellet was suspended in PBS. LEVs quantification was determined with bicinchoninic acid (BCA) assay protein assay kit. The size distribution and Zeta potential of the EVs were measured by dynamic light scattering (DLS). Particle concentration of LEVs were measured by nanoparticle tracking analysis (NTA). The surface morphology of the EVs was determined by TEM after negative staining with phosphotungstic acid.

4.4. Preparation of nanosized GRb1

To prepare for the GRb1, 1 mg of Rb1 was dissolved into 1 mL of PBS. The solution was stirred thoroughly to obtain an even dispersion of Rb1. Next, ultrasonic treatment at 480 W for a duration of 10 min was applied to promote GRb1 formation.

4.5. Preparation of GRb1@LEVs

GRb1@LEVs were prepared by quickly mixing lemon‐derived EVs and GRb1 at a weight ratio of 1:1, followed by sonication for 10 min and incubation at 37 °C for 1 h. The suspension was centrifuged at 100,000 g for 30 min at 4 °C to remove free GRb1 and obtain hybrid vesicles. For confocal laser scanning microscopy (CLSM), LEVs and GRb1 pre-labeled with DiD (excitation/emission = 644/665 nm) and DiO (excitation/emission = 484/501 nm), respectively, were used to prepare GRb1@LEVs and confocal laser scanning microscope was used to study the nanoparticles fusion. Förster resonance energy transfer (FRET) assay was also conducted to study the fusion process. Briefly, LEVs were doped with a FRET pair (DiI and DiD), into which GRb1 was added at different weight ratios (GRb1/LEVs = 0:1, 1:1, 4:1, and 10:1). The fluorescence spectrum of each sample was recorded between 550 and 750 nm at the excitation wavelength of 525 nm.

4.6. Synthesis of GRb1@LEVs-cRGD and its derivates

Preparation of GRb1@LEVs-cRGD: A convenient and rapid method was designed to endow the hybrid vesicles with lung‐targeting function. Briefly, GRb1@LEVs were incubated with cRGD-PEG2000-Chol (cRGD: LEVs = 4:1) at 4 °C under gentle stirring for 12 h, referring to established protocols [30,[57], [58], [59]]. Following this, GRb1@LEVs-cRGD were collected by centrifugation at 100,000×g for 2 h. The final suspension was stored at −80 °C until use. The size distribution, Zeta potential, and surface morphology of GRb1 and GRb1@LEVs-cRGD were performed using the same methods as described above.

Vanc/TIG remote-loaded GRb1@LEVs-cRGD: To prepare ammonium sulfate-contained vesicles, cRGD-PEG₂₀₀₀-Chol and was added at the specified feed ratio directly to GRb1@LEVs suspended in 300 mM ammonium sulfate. The ammonium sulfate gradient was formed by transferring the vesicles into PBS using a pre-equalized BeyoDesalt™ G-25 Mini Desalting Column. Next, Vanc or TIG at different drug inputs was incubated with the hybrid vesicles made with varying concentrations (0%, 2.5%, 5%, 7.5% and 10%) of cholesterol at 37 °C for 1 h. The unloaded free drug was removed using a BeyoDesalt™ G-25 Mini Desalting Column. The concentrations of Vanc and TIG were measured by their absorbance at 280 nm and 245 nm, respectively, using a microplate reader. Vesicles were first disrupted by acetonitrile and the precipitated proteins were removed by centrifugation, eliminating their associated background signal. The drug loading yield was determined as the mass of the loaded drug divided by the protein mass of the GRb1@LEVs-cRGD vesicles.

4.7. Cell culture

Human umbilical vein endothelial cells (HUVECs) and RAW264.7 (macrophage-like) cells were cultured in high-glucose Dulbecco's Modified Eagle Medium (DMEM) supplemented with 10% fetal bovine serum (FBS, v/v) and 1% penicillin-streptomycin solution (v/v). MH-S cells were maintained in complete RPMI-1640 medium supplemented with 10% fetal bovine serum (FBS, v/v) and 1% penicillin-streptomycin solution (v/v), while MLE-12 cells were cultured in the specific complete MLE-12 cells medium. All four cell lines were grown in a cell incubator at 37 °C with addition of 5% CO2.

4.8. In vitro cytotoxicity evaluation of GRb1@LEVs-cRGD

Cell Viability Assay: RAW264.7, HUVEC and MLE-12 cells were seeded into a 96-well plate at a density of 1 × 10⁴ cells per well and cultured for 12 h. The cells were then treated with LEVs, GRb1, GRb1@LEVs (equivalent GRb1 concentration) and GRb1@LEVs-cRGD (equivalent GRb1 concentration) at various concentrations for 24 h. Cell viability was assessed using the Enhanced Cell Counting Kit-8 (CCK8) according to the manufacturer's protocol.

Live/Dead Cell Staining: RAW264.7, HUVEC and MLE-12 cells were seeded in 12-well plates for 24 h. Then the culture medium was replaced with a fresh medium containing LEVs, GRb1, GRb1@LEVs and GRb1@LEVs-cRGD to incubate for 24 h. The untreated RAW264.7, HUVEC and MLE-12 cells were used as a control group. Following incubation, the cells were co-stained with Calcein-AM and Propidium Iodide (PI). Observations were made using a fluorescent inverted microscope, and the PI-positive rate was quantified using ImageJ software.

Cytoskeleton Staining: HUVECs were cultured into confocal Petri dishes until reaching 60%-70% confluence. Afterwards, DMEM solutions containing LEVs, GRb1, GRb1@LEVs and GRb1@LEVs-cRGD were added for 12 h or 48 h. Following the removal of the drug-containing medium, the cells were fixed with 4% paraformaldehyde solution for 10 min. Subsequently, the cells were permeabilized with 0.5% Triton X-100 and blocked with 5% BSA. At a final step, the cells were incubated with phalloidin in the dark overnight at 4 °C. Subsequently, the cells were treated with DAPI solution for 10 min, and fluorescence images were captured using a confocal microscope.

4.9. Cellular uptake of vesicles in vitro

RAW264.7 cells were seeded at a density of 3 × 10⁵ cells per well into confocal Petri dishes. Subsequently, a half of cells were exposed to 100 ng/mL of lipopolysaccharide (LPS), while the remaining served as controls. The cells were then treated with DiD-GRb1@LEVs and DiD-GRb1@LEVs-cRGD and cultured for 2 h. Then, the medium was removed, followed by washing three times with PBS. The cells were fixed with 4% paraformaldehyde for 15 min and permeabilized with 0.5% Triton X-100 for 10 min, blocked with blocking solution, incubated with phalloidin for 1 h, subsequently subjected to nuclear staining with DAPI for 10 min. Finally, the cells were imaged using 40 × oil immersion lens of Laser confocal microscope. Next, to assess the ability of GRb1@LEVs-cRGD to target HUVECs, the cells were cultured into confocal Petri dishes at a density of 2.5 × 10⁵ cells per well. The following experimental steps were consistent with the procedure described above and the cells without nanoparticles treatment were as control.

4.10. Evaluation of in vitro anti-inflammatory activity

MH-S cells and RAW264.7 cells were seeded in a 24-well plate at a density of 1 × 10⁵ cells per well, pretreated with PBS, LEVs, GRb1, GRb1@LEVs or GRb1@LEVs-cRGD for 2 h before the challenge with LPS (100 ng mL⁻¹). To ensure comparability, the concentration for all groups was standardized to 25 μg mL⁻¹ (based on GRb1 equivalent). 24 h later, the supernatants were collected and TNF-α, IL-6 contents were measured using ELISA kits.

Immunofluorescence (IF) staining was also employed to assess the ability of GRb1@LEVs-cRGD to inhibit the translocation of p65. RAW264.7 cells were seeded into confocal dish overnight, and then pretreated with the same standardized concentration (25 μg mL⁻¹) for 2 h before the challenge with LPS (100 ng mL⁻¹). 24 h later, Cells were subjected to p65 and DAPI immunofluorescence staining according to the protocol and then CLSM was employed to investigate the translocation of p65.

For RNA-seq and transcriptomic analysis, RAW264.7 cells were cultured as described above and treated with GRb1@LEVs-cRGD for 2 h, then challenged with LPS (100 ng mL⁻¹). Cells treated with PBS was regarded as control group. 6 h after addition of LPS, cells washed 3 times with PBS and then harvested with TRIzol and stored at −80 °C. The cells were sent to Shanghai OE Biotech Co., Ltd. for RNA-seq analysis and transcriptomic analysis.

4.11. Macrophage phenotype regulation in vitro

To evaluate M1/M2 macrophage polarization by flow cytometry. RAW 264.7 cells were first seeded in 6-well culture plates and incubated at 37 °C with 5% CO₂ to grow to 60%-70% confluence. After that, cells were subjected to the treatments described above and incubated with LPS (100 ng mL⁻¹) for 24 h. Subsequently, the cells were stained with PE-labeled anti-CD206 antibody and APC-labeled anti-CD86 antibody according to the manufacturer's instructions and then analyzed by flow cytometry.

4.12. Mitochondrial membrane potential assay

Mitochondrial membrane potential was evaluated using the JC-1 fluorescence probe. Briefly, cells cultured in confocal dishes were pre-treated with the same standardized concentration (25 μg mL⁻¹) for 24 h. Subsequently, the cells were then stained with JC-1 for 20 min at 37 °C and observed by fluorescent microscope.

4.13. Cell apoptosis analysis with flow cytometer

Cells were seeded in 6-well plates at a density of 6 × 10⁵ cells per well and cultured overnight. As previously mentioned, following treatment with the same standardized concentration (25 μg mL⁻¹), the cells were incubated with LPS (1 μg mL⁻¹) for 24 h. Subsequently, the cells were stained with annexin V and propidium iodide in annexin V binding solution at room temperature for 15 min. Apoptotic cells were quantified using flow cytometry, and the data were analyzed using FlowJo software.

4.14. Visualization of tight junctions

To evaluate the drug's effect on LPS-induced damage to endothelial integrity, HUVEC cells were individually seeded in 0.17 mm glass-bottom dishes until 100% confluence were reached. In the treatment groups, the same standardized concentration (25 μg mL⁻¹) were added to the culture dishes. Subsequently, the cells were stimulated with 1 μg mL⁻¹ LPS. After 24 h, IF staining was performed according to the aforementioned protocol, the cells were fixed with 4% paraformaldehyde PBS solution and blocked with 5% BSA for 1 h. Next, the cells were subsequently subjected to overnight incubation with ZO-1 antibody. The finalized samples were imaged using a confocal microscope.

4.15. Microdilution assay

Bacterial density was diluted to 5 × 10⁵ CFU mL⁻¹ in LB broth and added triplicate into individual wells of 96 well plate. Next, tested antibiotic (Vanc or TIG) or drug loaded vesicles were added performing 2-fold serial dilution. The plates were incubated at 37 °C for 16-18 h and measured OD₆₀₀ with microtiter plate reader. Wells with medium only and wells with no drugs represent background and control growth respectively. The minimum inhibitory concentration (MIC) refers to the lowest concentration of a drug that can inhibit bacterial growth as determined by visual reading and OD600 values.

4.16. Bacterial growth curves

Bacteria were cultured as methods above and the density was adjusted to 5 × 10⁵ CFU mL⁻¹. Different drugs combinations were added to bacterial culture. LB broth with no drug was determined as control. The absorbance of bacterial cultures at wavelength of 600 nm were detected at pre-determined time points (0, 2, 4, 6, 8, 12, 24 h). All tests were performed 3 times in triplicate on different days.

4.17. Animal tests

6–8-week-old male BALB/c mice weighting 22-25 g were purchased from Shanghai SLAC Laboratory Animal Co., Ltd. Mice were housed under specific pathogen-free (SPF) environments. With a 12 h cycle of light-dark, room temperature was held at 21-24 °C and the relative humidity was kept at 40%-60%. Animal experiments were conducted in accordance with the guidelines of ethics committee of Xinhua Hospital affiliated to Shanghai Jiao Tong University of Medicine. The license number of the experimental animal is XHEC-F-2025-063. Efforts were made to minimize animal suffering and reduce the number of animals used.

4.18. In vivo imaging system (IVIS)

To visualize in vivo distribution and inflammatory lung-targeting property of GRb1@ LEVs-cRGD, initially, mice were randomly assigned to three groups, two of which were administered LPS via intratracheal (i.t.) injection at a concentration of 4 μg mL⁻¹ following anesthesia, thereby establishing a model of acute lung injury. The other group was administered i.t. PBS as a control. Subsequently, mice were injected with Cy5.5-labeled GRb1@ LEVs or Cy5.5-labeled GRb1@ LEVs-cRGD (containing equal dose of cy5.5) via tail vein after 2 h. At 4 and 12 h post-intratracheal injection, major organs were harvested and imaged via IVIS spectrum.

4.19. Acute lung injury (ALI) animal model

An experimental model of ALI was induced in mice following a defined protocol. Mice were anesthetized via intraperitoneal injection of sodium pentobarbital (50 mg kg⁻¹) and positioned supine with their heads elevated to facilitate tracheal exposure. A dose of lipopolysaccharide (LPS, 4 mg kg⁻¹) was administered via tracheal dripping, and the mice were kept in the supine position for 5 min post-administration to ensure proper LPS distribution. After 1h, the mice were randomly divided into six groups (n = 5 per group). The therapeutic dosage was set at 8 mg kg⁻¹ (based on GRb1 equivalent) administered in 100 μL PBS via tail vein injection. This dosage was selected based on prior optimization and reported efficacy ranges for ginsenoside derivatives in acute lung injury models [60]. At 23 h post-treatment, mice were sacrificed, and orbital blood samples, bronchoalveolar lavage fluid (BALF) and lung tissues were collected for the further evaluation of therapeutic effects. Pulmonary edema in ALI mice was assessed by measuring the wet/dry weight ratio of the lungs. Lung tissues were fixed with 4% paraformaldehyde and sectioned for hematoxylin and eosin (H&E) staining and IF staining to assess inflammatory conditions and the permeability of endothelium and epithelium.

4.20. Mouse severe Kp pneumonia model

A severe bacterial pneumonia model was established in mice using Kp NDM. Mice were administered Kp NDM (10⁸ CFU mL⁻¹, 30 μL) intratracheally. After 1 h of infection, the mice were treated with PBS, free TIG, TIG/GRb1@ LEVs-cRGD (equivalent to 15 mg kg⁻¹ body weight of free TIG) via tail vein. At 23 h post-treatment, mice were sacrificed to evaluate the antibacterial effect. Lung tissues were homogenized and cultured on LB plates to determine bacterial burden through CFU counts. BALF supernatants and cell precipitates were collected separately by centrifugation. Cell precipitates were resuspended in 1 mL of PBS buffer for cell counting. The supernatant was used to measure the levels of total protein and inflammatory factors in BALF by BCA Protein Assay Kit. The remaining lung tissues were subjected to anti-inflammatory evaluation and pathological analysis.

4.21. Mouse MRSA pneumonia model

Mice were infected by MRSA (10⁸ CFU mL⁻¹, 30 μL) by intratracheal administration. After 1 h of bacterial infection, the mice received treatment with PBS, Vanc, Vanc/GRb1@ LEVs-cRGD (equal to 8 mg kg⁻¹ Vanc). After 23 h of treatment, the BALF and lungs were collected from mice after euthanasia for the assessment of antibacterial and anti-inflammatory effects.

4.22. Biocompatibility in vivo

To assess the safety of GRb1@LEVs-cRGD in vivo, mice were i.v. injected through tail vein with 100 μL GRb1@LEVs-cRGD (8 mg kg⁻¹) once a day. After 7 days of treatment, blood was collected from sacrificial mice for blood routine analysis, and biochemical assays. Major organs were sectioned for H&E staining to evaluate potential histopathological damage. Lung tissue was sectioned for TUNEL staining to assess apoptotic activity and ensure lung tissue safety.

4.23. Statistical analysis

Experiments were conducted in triplicate, ensuring independence between each trial. Results are expressed as mean ± standard deviation (SD). Data analysis was carried out utilizing GraphPad Prism 10.0 software. p < 0.05 was deemed statistically significant across all analyses, corresponding to a 95% confidence interval. Specifically, ∗ indicates p < 0.05, ∗∗ denotes p < 0.01, ∗∗∗ represents p < 0.001, and ∗∗∗∗ signifies p < 0.0001. The allocation of mice to different treatment groups and the selection of fields for microscopic inspection were performed randomly. Differences among the experimental groups were analyzed using a one-way ANOVA test and unpaired Student's T-test (two-tailed). The assessment of survival benefits was conducted using log-rank test.

CRediT authorship contribution statement

Xianhao Wu: Writing – original draft, Investigation, Formal analysis, Data curation, Conceptualization. Hanqing Li: Methodology, Data curation, Conceptualization. Huizhen Fan: Writing – original draft, Investigation. Anyi Wu: Validation, Formal analysis. Yingying Ma: Software, Resources. Li Wei: Methodology. Yun Wang: Resources. Yanfei Mao: Writing – review & editing, Visualization, Funding acquisition. Min Lu: Writing – review & editing, Resources, Funding acquisition, Conceptualization. Lai Jiang: Supervision, Resources, Project administration, Funding acquisition, Conceptualization.

Graphical content

Graphical content (Scheme 1, Fig. 1, Fig. 6, Fig. 7, Fig. 8a) was created with BioRender.com.

Ethics approval and consent to participate

All animal experiments were conducted in accordance with the guidelines of ethics committee of Xinhua Hospital affiliated to Shanghai Jiao Tong University of Medicine. The license number of the experimental animal is XHEC-F-2025-063.

Declaration of competing interest

The authors declare that no competing interests exist.

Appendix A. Supplementary data

The following is the Supplementary data to this article: