Chrysanthemum indicum L.-derived extracellular vesicles enhance the therapeutic efficacy of cyclosporine a against dry eye disease

Abstract

The pathological progression of dry eye disease (DED) involves a vicious cycle of oxidative stress and inflammation, posing a critical therapeutic challenge. Conventional therapies, such as cyclosporine A (CsA), are limited by poor corneal permeability and low ocular bioavailability. Here, we developed a novel, biocompatible nano-eye drop formulation using Chrysanthemum indicum L.-derived extracellular vesicles (CELNs) as a natural nanocarrier to engineer CsA-loaded CELNs (CsA@CELNs) for synergistic DED therapy. By combining CsA-mediated immunoregulation with enhanced corneal permeability and the intrinsic antioxidant and anti-inflammatory activities of CELN, the CsA@CELNs effectively disrupts the core pathogenic feedback loop of DED. In vitro and in vivo data demonstrated that a one-week, twice daily topical treatment with CsA@CELNs alleviated oxidative stress by scavenging reactive oxygen species (ROS) and activating the Nrf2/HO-1/NQO1 signaling pathway, while concurrently suppressing inflammation through inhibiting the NF-κB pathway and promoting macrophage repolarization from the pro-inflammatory M1 to the anti-inflammatory M2 phenotype. In a murine DED model, CsA@CELNs eye drop significantly restored tear secretion, promoted regeneration of corneal and conjunctival cells, and improved lacrimal gland histology. This multi-targeting CsA@CELNs system not only provides a safe and effective nanotherapeutic strategy for DED but also establishes plant-derived extracellular vesicles as a promising drug delivery platform for treating ocular surface and other inflammatory diseases.

Graphical Abstract

Supplementary Information

The online version contains supplementary material available at 10.1186/s12951-026-04402-5.

Background

Dry eye disease (DED) is an increasingly prevalent ocular disorder worldwide, driven by population-aging, environmental stressors, and lifestyle factors, including extended screen use [1]. DED is a multifactorial condition characterized by ocular symptoms and signs resulting from a loss of homeostasis of the tear film, often accompanied by ocular discomfort. It is associated with a complex interaction between reduced tear production, increased tear evaporation, and inflammation of the ocular surface [2]. Without timely intervention, the condition may progress to significant visual disturbances and impairment of quality of life.

Pathologically, oxidative stress, inflammatory responses, and tissue damage create a vicious cycle that mutually reinforce one another. Factors such as tear of hyperosmolarity and environmental stimuli induce excessive accumulation of reactive oxygen species (ROS) on the ocular surface, leading to oxidative damage and disruption of the corneal epithelial barrier. As critical signaling molecules, excess ROS activates inflammatory pathways such as NF-κB [3], promoting the release of pro-inflammatory cytokines. This not only triggers chronic inflammation but also drives the polarization of ocular surface macrophages toward the pro-inflammatory M1 phenotype, while suppressing the reparative M2-type macrophages, resulting in an imbalance of the immune microenvironment [4]. In return, this dysregulated state further exacerbates oxidative stress, directly leading to goblet cell apoptosis and tissue damage, forming the core mechanism underlying the persistent nature of DED [5].

Current clinical managements of DED focus on alleviating symptoms, reducing inflammation, and restoring tear film stability, with treatments ranging from lubricant eye drops and anti-inflammatory agents to more advanced immunomodulatory therapies [6]. Cyclosporine A (CsA), an immunosuppressant, has been an important treatment for moderate to severe DED by topical application [7]. CsA acts by inhibiting T-cell activation and the release of inflammatory cytokines, effectively targeting the underlying inflammatory process of DED, thereby improving the eyes ability to produce tears. However, the clinical success of CsA faces several challenges, with low drug delivery efficiency being particularly prominent. The inherent strong hydrophobicity of CsA makes it difficult to dissolve in aqueous eye drops. Additionally, the unique “lipid-aqueous-lipid” barrier structure of the cornea significantly restricts drug penetration after topical administration [8]. Furthermore, the short retention time on the ocular surface, due to rapid clearance from tears, further reduces the bioavailability, necessitating frequent administration. This not only reduces patient compliance but may also lead to irritative reactions due to the accumulation of preservatives in the eye drops [9]. These factors limit the full therapeutic potential of CsA, highlighting the urgent need to develop novel drug delivery systems to overcome the current treatment limitations.

Extracellular vesicles (EVs) are lipid bilayer-enclosed particles that are naturally released from both mammalian and plan cells. EVs function as crucial mediators of intercellular communication. They transport a diverse molecular cargo, including proteins, nucleic acids, lipids, and metabolites to recipient cells, thereby influencing their function [10]. In recent years, plant-derived extracellular vesicles (PDEVs) are gaining significant attention in the biomedical field because they exhibit notable therapeutic potential for treating various diseases, such as cancer and inflammatory disorders. A key advantage of PDEVs lies in their inherent safety profile, as they are generally biocompatible and pose a lower risk of immune reactions compared to vesicles from mammalian sources or synthetic nanoparticles. Furthermore, their plant origin allows for scalable and cost-effective production, making them a highly promising and sustainable platform for next-generation drug delivery and nanomedicine.

Chrysanthemum indicum L., a traditional Chinese medicinal herb, is known for its properties in clearing heat, detoxifying, and improving eye vision [11]. Modern pharmacological studies have revealed that extracts from this plant possess potent anti-inflammatory, antioxidant, and antimicrobial activities [12, 13]. Based on the above, we hypothesize that extracellular vesicles derived from Chrysanthemum indicum L. (named as CELNs) possess anti-inflammatory and antioxidant activities, as they carry the bioactive compounds inherent to the source plant. With nanoscale dimensions, the high surface aspect of CELNs allows them to seamlessly integrate into tear film formulations and be retained on the ocular surface, overcoming the rapid clearance that plagues many conventional eye drops. This facilitates prolonged contact with the compromised corneal and conjunctival epithelium. Furthermore, their lipid bilayer structure mimics biological membranes, promoting fusion and efficient intracellular delivery of their encapsulated anti-inflammatory and antioxidant cargo directly to the injured ocular surface cells. Unlike synthetic nanoparticles, CELNs are natural and inherently biocompatible, thereby minimizing the risk of ocular irritation. This delivery system will provide critical advantages for treating the sensitive and inflamed ocular surface in dry eye patients.

Based on the background mentioned above, this study proposes the use of CELNs as a delivery vesicle for CsA to construct a novel nano-drug formulation (CsA@CELNs). The system was formulated as an eye drop dosage form to systematically evaluate its therapeutic efficacy against DED. Research indicates that CELNs can encapsulate and deliver CsA efficiently, enhancing its corneal permeability and prolonging the ocular retention. Their intrinsic bioactive components can also synergize with CsA to exert antioxidant and anti-inflammatory effects, disrupting the vicious pathological cycle of DED. This study aims to provide experimental evidence for a synergistic treatment strategy based on natural vesicles, offering new perspectives for the clinical management of DED (Scheme 1).

Scheme 1.

Schematic illustration of the preparation of CsA@CELNs eye drop and its application in the treatment of DED. (A) Preparation workflow of CsA@CELNs eye drop. (B) Therapeutic mechanisms against DED encompass antioxidant, anti-inflammatory, and oxidative-damage-repair-promoting processes. (C) Characteristic changes of the ocular surface environment before and after treatment. (Created with BioRender.com)

Results and discussion

Preparation and characterization of CsA@CELNs

Flowers from fresh Chrysanthemum indicum plants were collected, washed with deionized water, and juiced using a juice extractor. Large particles were removed by low-speed centrifugation (3,500×g, 30 min), followed by medium-speed centrifugation (10,000×g, 1 h) to eliminate cellular debris. The supernatant was then filtered through a 0.8 μm aqueous filter membrane to remove plant fibers. Finally, the extract was subjected to high-speed centrifugation (150,000×g, 2 h), and the resulting pellet was collected and lyophilized to obtain CELNs (Fig. 1A). Transmission electron microscopy (TEM) characterization revealed that CELNs exhibited the typical “cup-shaped” morphology characteristic of PDEVs (Fig. 1B). Dynamic light scattering (DLS) analysis indicated the particle size of CELNs was 192.17 ± 1.9 nm with a polydispersity index (PDI) of 0.40 ± 0.02 (Fig. 1C). CELNs exhibited a zeta potential of − 30.80 ± 2.33 mV (Fig. 1D).

Fig. 1.

Characterization. (A) Schematic of CELNs isolation by ultrahigh-speed centrifugation (Created with BioRender.com). (B) Representative TEM image of CELNs (scale bar, 100 μm). (C) Hydration diameter distribution of CELNs. (D) Zeta potential of CELNs. (E) Representative TEM image of CsA@CELNs (scale bar, 100 μm). (F) Hydration diameter distribution of CsA@CELNs. (G) Zeta potential of CsA@CELNs. (H) FTIR spectra. (I) Particle-size stability of CsA@CELNs over 7 days. All the quantitative data are means ± SD; n = 3. (J) Protein stability of CsA@CELNs over 20 days. All the quantitative data are means ± SD; n = 3

Subsequently, CsA@CELNs were prepared using an ultrasonic method. TEM imaging showed a spherical morphology with intact membrane structure (Fig. 1E), with a reduced hydrodynamic diameter of 102.70 ± 1.39 nm and a notably narrower size distribution (PDI = 0.24 ± 0.04), suggesting that the ultrasonic treatment resulted in smaller and more homogeneous nanoparticles (Fig. 1F). Moreover, the moderate decrease in surface negative charge was indicated by the zeta potential value of − 26.67 ± 2.55 mV (Fig. 1G). To confirm the successful encapsulation of CsA, Fourier transform infrared (FTIR) spectroscopy was performed. The results showed that CsA@CELNs retained the characteristic peaks of CELNs, while the distinctive peaks of CsA at 2872.16, 1411.34, and 968.10 cm⁻¹ disappeared, confirming effective encapsulation of CsA within CELNs (Fig. 1H). High-performance liquid chromatography (HPLC) analysis determined an encapsulation efficiency of 70.58% and a drug loading capacity of 35.29% for CsA@CELNs. To further assess the stability, we incubated CsA@CELNs in DMEM medium supplemented with 10% fetal bovine serum. The particle size of CsA@CELNs remained unchanged after 7 days of storage at 4 °C (Fig. 1I), indicating good serum stability. Furthermore, no significant change in protein concentration was observed when lyophilized CsA@CELNs were stored sealed at 4 °C for 20 days (Fig. 1J), suggesting that the formulation possesses good short-term storage stability under refrigeration.

Preparation and characterization of CsA@CELNs eye drop

The CsA@CELNs eye drop was prepared using 1.0% sodium alginate and 5.0% glycerol as excipients. The resulting formulation exhibited a distinct Tyndall effect (Fig. 2A), indicating the presence of nanoparticles in the system. TEM result (Fig. 2B) clearly revealed the nanostructure of CsA@CELNs in the eye drops. DLS measurements showed a particle size of 110.93 ± 0.51 nm with a PDI of 0.21 ± 0.02 (Fig. 2E), suggesting a relatively uniform size distribution. The CsA@CELNs eye drop exhibited a pH of 7.4, thereby meeting the fundamental requirement for ophthalmic formulations by falling within the physiological pH range of tears (approximately 7.4–7.6). CsA release behavior was studied at 37 °C, pH 7.4 to mimic the temperature of ocular tissue. A cumulative release of (96.95 ± 4.87) % within 48 h was observed (Fig. 2C), indicating favorable drug release properties. To assess the stability, CsA@CELNs eye drop was sealed and stored at 4 °C. During the storage of 20 days, the protein concentration remained approximately unchanged (Fig. 2D). This finding indicated that CsA@CELNs maintained its integrity in eye drop formulation at the storage temperature. Rheological tests revealed that the CsA@CELNs eye drop exhibited nearly Newtonian fluid behavior, with apparent viscosity remaining stable under different shear rates (Fig. 2F). Therefore, their viscosity remains consistent under physiological shear conditions, such as blinking and instillation, facilitating uniform spreading on the ocular surface. SDS-PAGE analysis of the protein composition of CELNs showed distinct protein bands in the 23–32 kDa range (Fig. 2G). The protein profiles of both CsA@CELNs and the formulated eye drop were consistent with those of plain CELNs, indicating that the protein structure remained intact during drug loading and formulation.

Fig. 2.

Characterization of CsA@CELNs eye drop. (A) Appearance with Tyndall effect. (B) Representative TEM image of CsA@CELNs in the eye drop (scale bar, 50 μm). (C) CsA release from CsA@CELNs eye drop at pH 7.4. (D) Protein concentration in CsA@CELNs eye drop over 20 days. (E) Hydration diameter distribution of CsA@CELNs in the eye drop. (F) Rheological properties. (G) SDS-PAGE protein analysis. (H) Schematic of corneal flat-mount preparation (Created with BioRender.com). (I) Evaluation of corneal residence time (scale bars, 1000 μm). (J) Evaluation of corneal penetration depth (scale bars, 50 μm). All the quantitative data are means ± SD; n = 3

The corneal barrier is characterized by a multilayer of “lipid-aqueous-lipid” structure, which limits the penetration of conventional lipophilic eye drops. Additionally, rapid clearance by tear fluid further reduces bioavailability [14]. In this study, CELNs were employed as a carrier to overcome these challenges. Their amphiphilic structure promotes drug penetration across corneal layers [15], while their nanoscale size and surface adhesive proteins improve drug retention and diffusion within corneal tissue. Collectively, these attributes significantly enhance ocular drug absorption and therapeutic efficacy [16]. To investigate this mechanism, coumarin-6 (C6), a lipophilic fluorescent model compound, was used in place of CsA. While CELNs were labeled with the fluorescent dye 1,1’-Dioctadecyl-3,3,3’,3’-tetramethylindocarbocyanine perchlorate (DiI) to construct C6@CELNs-DiI nanoparticles for in vitro evaluation of corneal penetration and retention. As shown in Fig. 2H, mouse corneas were mounted on glass slides after administration. Fluorescence microscopy of mouse corneas revealed that both DiI and C6 signals were present throughout the entire cornea within 1 h and persisted for up to 12 h (Fig. 2I). Figure 2J further confirmed effective penetration across all corneal layers, with detectable fluorescence still evident at 12 h. Together, these results demonstrated that the C6@CELNs-DiI nanoparticles exhibited strong corneal permeability and sustained retention.

Cellular uptake and cytotoxicity

As naturally derived plant EVs, CELNs inherently encapsulate proteins, polysaccharides, nucleic acids, and other bioactive components [17]. Leveraging their carrier properties, we successfully constructed the CsA@CELNs drug delivery system. The release of CsA and other bioactive ingredients highly depends on the effective internalization of the EVs by target cells. Thus, cellular uptake of C6 labelled CELNs (C6@CELNs) on HCE-T and RAW264.7 cells was examined. Both HCE-T (Fig. 3A and B) and RAW264.7 cells (Fig. 3D and E) took up C6@CELNs in a time-dependent manner, reaching peak intracellular accumulation at 8 h after incubation. Then the fluorescence signal gradually decreased, indicating the gradual digestion and metabolism of C6@CELNs within the cells. Fluorescence microscopy observations revealed strong colocalization of the green fluorescence from the nanoparticles and the red fluorescence from the lysosomal marker LysoTracker, indicating that CELNs primarily enter cells via endocytosis and are processed through the lysosomal pathway. The acidic environment and abundant hydrolytic enzymes in lysosomes degrade the vesicular structure, thereby releasing the encapsulated drug and bioactive components to exert subsequent biological effects [18].

Fig. 3.

Analysis of cellular uptake and cytotoxicity. (A) Uptake of CsA@CELNs by HCE-T cells (scale bars, 100 μm). (B) Quantification of fluorescence intensity in Fig. 3A. (C) Cytotoxicity evaluation of CELNs and CsA@CELNs on HCE-T cells. (D) Uptake of CsA@CELNs by RAW264.7 cells (scale bars, 100 μm). (E) Quantification of fluorescence intensity in Fig. 3D. (F) Cytotoxicity evaluation of CELNs and CsA@CELNs on RAW264.7 cells. All the quantitative data are means ± SD; n = 3. (^**P < 0.01, ^****P < 0.0001)

To evaluate the cellular safety, we assessed the viability of HCE-T and RAW264.7 cells treated with different concentrations of CELNs and CsA@CELNs using the CCK-8 assay. As shown in Fig. 3C, the viability of HCE-T cells treated with CELNs remained close to 100% even at a concentration of 800 µg/mL, indicating no significant cytotoxicity of CELNs. However, noticeable cytotoxicity was observed when the concentration of CsA@CELNs exceeded 500 µg/mL. This may be attributed to the cytotoxic CsA loaded in CELNs. The impact of free CsA on HCE-T cell viability was concentration-dependent (Figure S1). While viability remained above 90% at 5 µg/mL, it induced significant cell death at higher concentrations. For RAW264.7 cells, treatment with both CELNs and CsA@CELNs at 100 µg/mL resulted in cell viability close to 100% (Fig. 3F), demonstrating good biocompatibility of the formulations at this concentration. In contrast, free CsA exhibited significant toxicity toward these cells at concentrations exceeding 5 µg/mL (Figure S2).

In vitro antioxidant effects, cellular oxidative damage repair, and mechanism investigation

Oxidative stress is a key pathological process in the development and progression of DED, characterized by the excessive accumulation of ROS, which leads to disruption of the corneal epithelial barrier and activation of inflammation [19]. As the outermost cells of the cornea, corneal epithelial cells not only play a central role in maintaining the ocular surface physical barrier but are also actively involved in innate immune responses. Under oxidative stress, these cells can release various inflammatory cytokines and activate intracellular stress signaling pathways, thereby exacerbating the pathological progression of DED. In this study, hydrogen peroxide (H₂O₂) was used to establish an oxidative damage model in HCE-T cells to simulate the high oxidative stress microenvironment of DED. Based on this model, the therapeutic potential of CsA@CELNs on oxidatively damaged cells was evaluated (Fig. 4A). Treatment with 250 µM H₂O₂ for 1 h reduced cell viability to 50% (Figure S3). Under this condition, CsA, CELNs, and CsA@CELNs were subsequently incubated for 24 h. Within the concentration range of 0–5 µg/mL, the CsA group showed no significant change in cell viability (Fig. 4B). In contrast, CELNs effectively promoted the repair of oxidative damage in HCE-T cells, with viability rising to (88.52 ± 9.04)% at 20 µg/mL and the protective effect extending to concentrations as high as 200 µg/mL (Fig. 4C). Similarly, CsA@CELNs also demonstrated significant reparative efficacy at 20 µg/mL (Fig. 4D). We measured intracellular ROS level in HCE-T cells using the 2’,7’-dichlorodihydrofluorescein diacetate (DCFH-DA) probe. H₂O₂ stimulation successfully induced oxidative stress, increasing ROS levels to 2.29 times that of the normal group. Treatment with CsA and CELNs alone produced moderate reductions of 20.86% and 30.62%, respectively. However, the CsA@CELNs formulation was markedly more effective, achieving a 41.44% reduction in ROS and representing the most significant inhibitory effect (Fig. 4E and F).

Fig. 4.

In vitro antioxidant efficacy and mechanism of CsA@CELNs. (A) Schematic illustration of the construction and treatment of the HCE-T oxidative damage model (Created with BioRender.com). Cytotoxic effects of (B) CsA, (C) CELNs, and (D) CsA@CELNs on HCE-T cells. (E) Representative fluorescence images of ROS generation in HCE-T cells detected using a DCF fluorescent probe (scale bar: 100 μm). (F) Quantitative analysis of the mean DCF fluorescence intensity measured by flow cytometry. (G) MDA analysis. (H) SOD analysis. (I) Representative fluorescence images of mitochondrial damage in HCE-T cells assessed using the JC-1 probe (scale bar: 100 μm). (J) Representative immunofluorescence images of Nrf2 in HCE-T cells (scale bar: 100 μm) and the fluorescence intensities of Nrf2 and DAPI in HCE-T cells. (K) Quantification of fluorescence intensity in Fig. 4J. All the quantitative data are means ± SD; n = 3. (^*P < 0.05, ^**P < 0.01, ^***P < 0.001, ^****P < 0.0001)

Since ROS accumulation is closely linked to mitochondrial dysfunction [20], excessive ROS exacerbates this condition through oxidative damage to mitochondrial membranes, proteins, and DNA. Consequently, this impairment disrupts energy metabolism and triggers apoptosis [21]. To evaluate mitochondrial functional status, we employed the JC-1 fluorescent probe to measure mitochondrial membrane potential. JC-1 forms aggregates under normal membrane potential but dissociates into monomers when the membrane potential decreases. As shown in Fig. 4I, untreated cells predominantly exhibited red fluorescence (JC-1 aggregates), indicating intact mitochondrial structure [22]. In contrast, the H₂O₂ model group displayed the most prominent green fluorescence, suggesting severe ROS-induced mitochondrial damage. Compared to the model group, all drug-treated groups showed weakened green fluorescence and enhanced red fluorescence, indicating varying degrees of mitigation in mitochondrial damage. Notably, the CsA@CELNs group exhibited the strongest red fluorescence, demonstrating the most potent restorative effect.

Superoxide dismutase (SOD) is a key antioxidant enzyme responsible for clearing superoxide free radicals, and its activity directly reflects the cellular antioxidant capacity [23]. Malondialdehyde (MDA), as an end product of lipid peroxidation, is widely used to assess the extent of oxidative damage [24]. Thus, the antioxidant effects of CsA@CELNs were further systematically evaluated by measuring SOD activity and MDA content in cells from all treatment groups. As shown in Fig. 4G, H and H₂O₂ stimulation significantly decreased SOD activity and markedly increased MDA levels, indicating impaired antioxidant function and exacerbated lipid peroxidation under oxidative stress. Treatment reversed the oxidative stress profile observed in the model group, resulting in elevated SOD activity and reduced lipid peroxidation as measured by MDA. This effect was most robust in the CsA@CELNs group, which demonstrated a 110.8% boost in SOD and a 59.9% drop in MDA. The results demonstrated that CsA@CELNs mitigate oxidative stress by scavenging ROS, enhancing endogenous antioxidant enzymes, and alleviating lipid peroxidation. Distinct from simple plant extracts, the vesicular structure of CELNs protects their bioactive cargo and facilitates efficient intracellular delivery, thereby effectively restoring the redox homeostasis essential for corneal epithelial survival.

The antioxidant effects of cells typically rely on the activation of endogenous defense systems. To investigate whether the drug activates these systems, we analyzed the subcellular localization of the central antioxidant regulator erythroid 2-related factor 2 (Nrf2). Under basal conditions, Nrf2 binds to Keap1 and remains localized in the cytoplasm [25]. Oxidative stress or drug treatment triggers its nuclear translocation, initiating the expression of a series of downstream antioxidant genes, including heme oxygenase-1 (HO-1) and NAD(P)H: quinone oxidoreductase 1(NQO1) [26]. As shown in Fig. 4J, K and H₂O₂ stimulation partially activated Nrf2 nuclear translocation in HCE-T cells. Quantitative analysis revealed that the fluorescence intensity of Nrf2 in the model group was 2.94 times higher than the normal group. All drug treatments further enhanced Nrf2 fluorescence intensity, with the CsA, CELNs, and CsA@CELNs groups exhibiting 1.60-fold, 1.80-fold, and 2.89-fold increases compared to the model group, respectively. Among them, the most significant nuclear accumulation of Nrf2 was observed in CsA@CELNs treatment group, suggesting the activation of endogenous antioxidant defense pathway.

In vitro anti-inflammatory effects and mechanism investigation

Macrophages play a critical role in the development and progression of DED, not only by initiating and modulating ocular surface inflammation but also by influencing disease progression through phenotypic polarization. In the pathological microenvironment of DED, macrophages are predominantly activated into the pro-inflammatory M1 phenotype, which highly expresses inflammatory factors such as TNF-α, IL-1β, and iNOS, exacerbating inflammatory damage, apoptosis, and decreased tear film stability in corneal and ocular surface tissues [27]. In contrast, M2 macrophages, which possess anti-inflammatory and reparative functions, promote tissue repair and inflammation resolution through the secretion of factors such as IL-10 and TGF-β [28]. In recent years, regulating macrophage polarization has emerged as a novel therapeutic strategy for DED. Promoting repolarization from the M1 to the M2 phenotype can not only suppress excessive inflammatory responses but also help restore homeostasis of the ocular surface microenvironment, offering new perspectives for immunomodulatory therapy of DED. Based on this strategy, an M1 macrophage model was established by lipopolysaccharide (LPS) induction in RAW264.7 cells to mimic inflammation in DED, thereby investigating the effect of CsA@CELNs on macrophage repolarization to the M2 phenotype (Fig. 5A).

Fig. 5.

In vitro anti-inflammatory efficacy and mechanism of CsA@CELNs. (A) Schematic illustration of the construction and treatment of the RAW264.7 inflammation model (Created with BioRender.com). (B) NO release measurement. (C) Expression of CD86⁺ and CD206⁺ on RAW264.7 macrophages determined by flow cytometry. (D) M1/M2 phenotype ratio calculated from Fig. 5C. (E) Representative immunofluorescence images of NF-κB p65 in RAW264.7 cells (scale bar: 100 μm). (F) Levels of TNF-α, IL-1β, IL-6, and IL-10 in cells measured by ELISA (n = 3). All the quantitative data are means ± SD; n = 3. (^*P < 0.05, ^**P < 0.01, ^***P < 0.001, ^****P < 0.0001)

Nitric oxide (NO) serves as a classical marker of M1 macrophage activation, and its expression level directly reflects the degree of M1 polarization and related inflammatory status [29]. We measured the NO content in the cell supernatant after treatments. As shown in Fig. 5B and Figure S4, after 24 h of incubation with 100 ng/mL LPS, the NO release from RAW264.7 cells increased to 12.18 times that of the normal group. Within the concentration range of 1–200 µg/mL, both CELNs and CsA@CELNs significantly inhibited NO release. At the same concentration, CsA@CELNs exhibited a stronger inhibitory effect, which may be attributed to the loaded CsA component, as CsA itself also can suppress NO release (Figure S5). The most significant inhibition of NO was observed at the concentration of 20 µg/mL of CsA@CELNs, which was therefore selected for further study.

To evaluate whether M1 macrophages underwent repolarization, we used CD86 and CD206 antibodies to label M1 and M2 macrophages, respectively, and measured their fluorescence intensity by flow cytometry [30]. LPS stimulation significantly increased the number of M1 macrophages, while the number of M2 macrophages showed no significant change (Fig. 5C and D). All treatments significantly lowered the M1/M2 ratio from the model group’s value of 2.36, a shift driven by a decrease in M1 and an increase in M2 macrophage populations. The CsA, CELNs, and CsA@CELNs groups achieved ratios of 1.12, 0.97, and 0.64, respectively. Notably, the CsA@CELNs group’s ratio of 0.64 was nearest to the normal baseline of 0.77, indicating it had the most potent effect on normalizing macrophage polarization.

The NF-κB signaling pathway plays a central regulatory role in inflammatory responses, and the nuclear translocation of its p65 subunit is a critical step in pathway activation [31]. Activated p65 enters the nucleus and promotes the transcription of inflammatory cytokines such as TNF-α, IL-1β, and IL-6, thereby amplifying the inflammatory response [32]. Sustained activation of this pathway is closely associated with excessive inflammation in immune cells, including macrophages. To evaluate the inhibitory effect of the formulation on NF-κB activation, we examined the subcellular localization of p65 in RAW264.7 cells using immunofluorescence. As shown in Fig. 5E, LPS stimulation significantly induced p65 nuclear translocation, showing the activation of the inflammatory pathway [33]. The inhibition induced by CsA is consistent with its established pharmacological action [34], whereas the effect of CELNs may be attributed to nanocarrier-mediated modulation of the intracellular environment. By integrating both mechanisms, CsA@CELNs achieved the strongest inhibitory effect. Notably, the nuclear fluorescence intensity in this group was nearly restored to normal levels, indicating CsA@CELNs suppress NF-κB signaling activation effectively and exerts anti-inflammatory effects. We further measured the expression levels of key inflammatory cytokines TNF-α, IL-1β, IL-6, and the anti-inflammatory cytokine IL-10 in the supernatant of RAW264.7 cells after drug treatment using ELISA kit. Compared with the normal group, LPS stimulation significantly increased the release of TNF-α, IL-1β, and IL-6, while no significant change was observed in IL-10 expression (Fig. 5F). After treatments, all formulations demonstrated a bidirectional regulatory effect on cytokines. Both CsA and CELNs effectively suppressed the release of pro-inflammatory factors (TNF-α, IL-1β, and IL-6) and promoted the production of the anti-inflammatory cytokine IL-10, confirming that each possesses distinct anti-inflammatory activity. Importantly, CsA@CELNs treatment also markedly up-regulated the expression of IL-10, outperforming the single-component treatment groups. Rather than serving merely as passive carriers, CELNs function as bioactive modulators targeting the NF-κB axis. By facilitating efficient internalization into macrophages, they actively reprogram the immune microenvironment, thereby amplifying the therapeutic efficacy of the loaded CsA. These results demonstrated that CsA@CELNs possess multi-faceted anti-inflammatory effects through synergistic modulation of macrophage polarization and inhibition of the NF-κB pathway.

In vivo therapeutic effects on DED

To evaluate the therapeutic efficacy of CsA@CELNs eye drop on DED, this study established a mouse dry eye model by topical administration of 0.2% benzalkonium chloride (BAK). As a common ophthalmic preservative, BAK is frequently used as an inducer for dry eye models due to its ability to disrupt corneal epithelial integrity, reduce tear film stability, and induce excessive generation of ROS, leading to oxidative stress and inflammatory responses [35]. Starting from day 8 of modeling, the control and model groups were treated with saline eye drops, while the treatment groups were administered with lkervis^® (the commercial 0.1% CsA eye drop), CELNs, and CsA@CELNs eye drop, respectively. All groups were treated twice daily (at 8 am and 4 pm) for 7 days (Fig. 6A). Before treatment, mice in the model group exhibited typical dry eye symptoms, including ocular surface dryness, uneven tear distribution, corneal defects, and increased secretions, indicating successful model establishment [36].

Fig. 6.

In vivo treatment of BAK-induced DED. (A) Schematic diagram of the dry eye model establishment and treatment protocol (Created with BioRender.com). (B) Representative images of sodium fluorescein staining on the mouse ocular surface. (C) Statistical scoring of sodium fluorescein staining on the ocular surface. (D) Tear secretion volume (length of wetting of the phenol red thread). (E) Eyeball weight. (F) Body weight changes. (G) Representative H&E sections of corneal tissue (scale bar: 50 μm). (H) PAS-stained conjunctival epithelial tissue (red triangles indicate goblet cells) (scale bar: 200 μm). (I) Representative sections of lacrimal gland tissue (blue arrows indicate lymphocyte infiltration, red arrows indicate pyknosis of glandular epithelial cells) (scale bar: 50 μm). (J) and (K) represent the thickness of the cornea and the corneal epithelium, respectively. (L) Number of goblet cells in the conjunctiva. (M) Number of lymphocytes in the lacrimal gland. All the quantitative data are means ± SD; n = 3. (*P < 0.05, **P  < 0.01, ***P < 0.001, ****P < 0.0001)

We then used sodium fluorescein staining to assess the degree of corneal epithelial damage. This method relies on the binding of fluorescein to damaged areas of the cornea, which produces green fluorescence under cobalt blue light excitation, visually reflecting the disruption of the epithelial barrier [37]. As shown in Fig. 6B, the corneas of mice in the normal group showed almost no fluorescence staining, while the model group exhibited extensive and intense fluorescence on the ocular surface, indicating severe epithelial damage. However, the extent of fluorescence staining was reduced after treatments. According to the 10-point scoring system (a score > 6.0 indicates severe dry eye symptoms), the CsA and CELNs groups still showed fluorescent spots and higher scores after 7 days of treatment, suggesting incomplete repair of corneal damage. In contrast, the CsA@CELNs group demonstrated significantly lower scores, with corneal staining conditions similar to those of the normal group (Fig. 6C), providing preliminary evidence of its significant therapeutic effect on DED. As shown in the Figure S6, ocular surface photographs provided a direct comparison among the groups. Mice in the normal group exhibited clear, transparent corneas without conjunctival hyperemia. In contrast, the model group showed marked deterioration, characterized by rough, lusterless corneas and significant conjunctival hyperemia. All treatments induced measurable recovery from this damaged state. Both CsA and CELNs monotherapies increased corneal smoothness and reduced hyperemia. The CsA@CELNs group, however, exhibited the most substantial restoration, with corneal transparency and conjunctival health nearly matching the normal state, thereby visually confirming its superior efficacy in promoting ocular surface repair. Subsequently, we measured the tear secretion volume of mice in each group after treatment using the Schirmer tear test (Fig. 6D). The wetting length in the normal group was (6.70 ± 0.56) mm, while it significantly decreased to (2.13 ± 0.21) mm in the model group, indicating markedly reduced tear secretion in dry eye model mice. Due to low drug solubility and short corneal retention time, the CsA group exhibited a wetting length of (5.23 ± 0.42) mm. In contrast, the CsA@CELNs group showed significant recovery in tear secretion after treatment, achieving a wetting length of (6.20 ± 0.46) mm, which was comparable to the normal group. These results suggested that the CELNs carrier effectively increased the bioavailability, thus enhancing the therapeutic efficacy of CsA.

Notably, the eyeball weight of mice in the model group was significantly higher than that in the normal control group (Fig. 6E), which may be closely associated with BAK-induced ocular surface inflammatory responses and tissue edema. Previous studies have indicated that BAK, as an irritant preservative, can disrupt the blood-aqueous barrier, increase vascular permeability, and cause local fluid exudation, potentially contributing to the increase in eyeball weight [37]. In contrast, the wet weights of eyeballs in all treatment groups showed no significant differences compared to the normal group. Additionally, no significant changes in body weight were observed in any group after treatment compared to baseline (Fig. 6F), preliminarily indicating that the dry eye modeling method and treatments with CsA, CELNs, and CsA@CELNs did not induce significant systemic toxicity or adverse reactions.

We further analyzed structural changes in the ocular tissue after treatment through histological evaluation. Hematoxylin and eosin staining (H&E) results (Fig. 6G) revealed that in the normal group, corneal cells were tightly and orderly arranged with no inflammatory cell infiltration. In contrast, the model group exhibited structural damage to the corneal epithelium, characterized by thinning of the epithelial layer, reduced cell density (Fig. 6J-K), loosening of the stromal layer, and the presence of numerous vacuoles and inflammatory cell infiltration (Fig. 6M). Moreover, the therapeutic outcomes varied markedly among the treatment groups. CsA administration significantly alleviated corneal structural damage and reduced inflammatory cell infiltration, consistent with its well-established anti-inflammatory properties. The CELNs group also demonstrated robust efficacy, confirming that the nanocarrier itself possesses intrinsic bioactivity capable of promoting repair and suppressing inflammation. Notably, the CsA@CELNs group achieved the most comprehensive outcomes, characterized by the most complete corneal structural recovery and the lowest level of inflammatory infiltration. This superior structural and anti-inflammatory recovery demonstrated the enhanced therapeutic efficacy of the CsA@CELNs formulation.

In the development and progression of DED, conjunctival goblet cells often exhibit a reduction in number and functional impairment. These cells secrete mucins, forming the core component of the tear film’s mucus layer, which plays a key regulatory role in the adhesion, spreading, and stability of tears on the ocular surface [38]. To evaluate the effects of different treatment regimens on goblet cells, periodic acid-Schiff (PAS) was used to stain conjunctival tissue. By specifically labeling intracellular mucins, we quantitatively analyzed changes in goblet cell density and morphology (Fig. 6H and L). After induction of the dry eye model with BAK, conjunctival tissue analysis revealed severe goblet cell atrophy and a statistically significant reduction in cell density in the model group compared to healthy controls. While subsequent treatments with various formulations initiated a degree of cellular recovery, CsA@CELNs treatment prompted a potent recovery. This group displayed the highest goblet cell density which was significantly greater than both the model group and the other treatment groups. Furthermore, qualitative morphological analysis revealed that the goblet cells in the CsA@CELNs group predominantly retained a healthy, plump, and circular architecture.

The structural and functional integrity of the lacrimal gland is essential for maintaining ocular surface moisture, as this gland is the primary source of the tear film’s aqueous layer [39]. As shown in Fig. 6I, the lacrimal glands of normal mice exhibited intact structure, with tightly arranged and regularly shaped acinar cells, and only a small number of lymphocytes in the interstitium. However, the model group presented with severe pathology, including loosely arranged acini and widespread atrophy. The increased vacuoles within acinar cells, along with extensive lymphocyte infiltration were also observed, confirming local inflammation. Therapeutic intervention successfully mitigated these pathological changes, with all treatment groups displaying restored glandular architecture and a significant reduction of lymphocytic infiltration.

In vivo underlying mechanisms

Previous studies have indicated that the BAK-induced dry eye model can trigger corneal epithelial cell apoptosis, thereby accelerating the progression of disease [40]. To evaluate the apoptosis and proliferation status of corneal cells, we performed TUNEL and Ki67 staining assays after treatments. Histological evaluation (Fig. 7A) identified a pronounced imbalance in the corneal tissue of the model group, with widespread TUNEL-positive apoptosis and scant Ki67-positive proliferation, confirming a state of severe damage with inadequate repair. Following treatment, all groups exhibited a marked reversal of this balance, characterized by a significant decrease in apoptosis and a concurrent increase in cellular proliferation. In the CsA@CELNs treated group, strong proliferation could be observed, indicating the enhanced corneal repair capacity.

Fig. 7.

Therapeutic mechanism of CsA@CELNs. (A) Representative TUNEL and Ki-67 staining of corneal tissue (scale bar: 50 μm). (B) Schematic illustration of the mechanism underlying the therapeutic effect of CsA@CELNs on DED (Created with BioRender.com). Western Blot analysis of Nrf2 protein expression in (C) Cytoplasmic and (D) nuclear fractions of ocular tissue (n = 3). (E) SOD production (n = 5). (F) MDA production (n = 5). (G) Representative immunohistochemical results of HO-1 and NQO1 in the cornea, along with ROS production in corneal tissue (scale bar: 50 μm, 50 μm, 50 μm). (H) Western Blot analysis of NF-κB p65 and phosphorylated NF-κB p65 (p-p65) proteins in ocular tissue (n = 3). (I) Representative images of iNOS production in corneal tissue (scale bar: 50 μm). (J) Expression of F4/80⁺, CD86⁺, and CD206⁺ cells in corneal tissue (scale bar: 50 μm). (K) Expression of CD4⁺T cells in corneal tissue (scale bar: 50 μm). (L) Levels of TNF-α, IFN-γ, IL-1β, IL-6, and IL-10 in corneal tissue after 7 days of treatment (n = 5). (M) Western Blot analysis of MMP-3/9 protein expression (n = 3). All the quantitative data are means ± SD. (*P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001)

The core pathology of DED stems from a dysregulation of the ocular surface, where oxidative stress, persistent inflammation, and corneal structural damage disrupt homeostasis. Stressors such as hyperosmolar tears and mechanical friction can induce the production of ROS in epithelial cells, triggering oxidative stress [41]. Excessive ROS not only damages cells directly but also acts as an amplifier of inflammation, activating the NF-κB inflammatory pathway and promoting the expression of pro-inflammatory cytokines such as TNF-α, IL-6, IL-1β, and IFN-γ, which recruit and activate more inflammatory cells, forming a vicious cycle [42]. In this process, the imbalance in macrophage polarization is particularly critical: overactivation of M1 macrophages exacerbates inflammation and tissue damage, while the relative deficiency of reparative M2 macrophages contributes to sustained ocular surface inflammation and impaired healing [43]. Concurrently, the inflammatory microenvironment driven by NF-κB promotes the activation of CD4⁺ T cells, which secrete high levels of IFN-γ. This pro-inflammatory cytokine aggravates corneal epithelial injury and impedes tissue repair [44] (Fig. 7B). Moreover, a reciprocal crosstalk between M1 macrophages and CD4⁺ T cells sustains and amplifies local inflammation, whereas M1 macrophages enhance T cell activation, while IFN-γ from T cells reinforces M1 polarization. This positive feedback loop perpetuates the inflammatory cascade, ultimately exacerbating corneal barrier dysfunction and disease progression [45].

The Nrf2/ARE signaling pathway plays a central role in regulating oxidative stress and inflammation [46]. As a key antioxidant transcription factor, the nuclear translocation of Nrf2 directly regulates the expression of downstream antioxidant enzymes such as HO-1 and NQO1, and influences intracellular inflammatory processes [47, 48]. Western Blot results (Fig. 7C and D) showed that both cytoplasmic and nuclear Nrf2 levels were low in the normal group, indicating homeostatic regulation under physiological conditions. The model group exhibited a slight increase in nuclear Nrf2, suggesting stress-induced nuclear translocation. Both CsA and CELNs treatments showed enhanced Nrf2 nuclear translocation, along with inhibited degradation of cytoplasmic protein, indicating that both CsA and CELNs promoted Nrf2 protein stabilization and nuclear migration. CsA@CELNs treatment showed the highest accumulation of Nrf2 in both the cytoplasm and nucleus, suggesting that this formulation combining CsA and CELNs efficiently activates the Nrf2 pathway.

To further evaluate the downstream antioxidant effects, we examined the expression of HO-1 and NQO1 in corneal tissue through immunohistochemistry. Both proteins exhibited low expression level in the model group (Fig. 7G), while the CsA monotherapy showed significantly enhanced expression, with positive signals mainly distributed in the epithelial and stromal layers. After CELNs and CsA@CELNs treatments, the expression of HO-1 and NQO1 recovered to near-normal levels. This recovery indicated an effective alleviation of oxidative stress and restoration of antioxidant enzyme homeostasis, an effect attributable to the potent bioactive components within the CELNs. Consistent with this, ROS fluorescence staining revealed a significant increase in fluorescence signal in the model group, which was markedly reduced in the CsA@CELNs group (Fig. 7G). Biochemically, CsA@CELNs reversed the oxidative stress profile by significantly increasing SOD activity and decreasing MDA content compared to the model group, which further confirmed that this formulation enhanced antioxidant capacity and suppressed lipid peroxidation (Fig. 7E and F).

Given the clear antioxidant effects, we further investigated the regulatory role of CsA@CELNs on inflammatory pathways. NF-κB is a central hub in DED-related inflammation, and its activation state is closely associated with macrophage polarization and pro-inflammatory cytokine secretion [49]. In the DED model, a significant increase in the p-p65/p65 level was observed in Western Blot results (Fig. 7H, Figure S7), indicating NF-κB activation. All treatments mitigated this activation, but the inhibitory effect was strongest in the CsA@CELNs group, followed by the CsA and CELNs monotherapies. Then, we performed immunofluorescence staining to clarify macrophage polarization status. Analysis revealed a pro-inflammatory state in the model group, with elevated M1 (F4/80⁺CD86⁺) and diminished M2 (F4/80⁺CD206⁺) populations. CsA and CELNs treatments promoted a shift toward the M2 phenotype. CsA@CELNs significantly reduced total macrophage infiltration, as indicated by a substantial decrease in overall F4/80⁺ intensity, thereby highlighting its superior anti-inflammatory efficacy (Fig. 7J). Meanwhile, immunofluorescence results revealed substantial infiltration of CD4⁺ T cells in the corneal stroma of the model group. Following treatment with CsA@CELNs, the fluorescence intensity of CD4⁺ T cells was markedly reduced, suggesting that this formulation can inhibit T cell migration to the ocular surface and decrease the release of local inflammatory factors, thereby further attenuating the inflammatory amplification cascade (Fig. 7K).

The pro-inflammatory state was further confirmed at the molecular level. The model group exhibited intense inducible nitric oxide synthase (iNOS) fluorescence, a classic M1 marker, which was markedly reduced by CsA@CELNs treatment (Fig. 7I). This shift was corroborated by cytokine analysis. Pro-inflammatory cytokines (TNF-α, IL-6, IL-1β, IFN-γ) were significantly reduced by CsA@CELNs, whereas anti-inflammatory cytokine IL-10 was markedly increased (Fig. 7L). These results indicated that CsA@CELNs alleviated DED by normalizing the dysregulated immune-inflammatory microenvironment.

Previous research showed that activated CD4⁺ T cells upregulate the expression of MMP-3 and MMP-9 through secreting IFN-γ and IL-17 [34]. These proteases play critical roles in extracellular matrix (ECM) degradation and inflammatory regulation, with their expression levels reflecting the balance between corneal tissue damage and repair [50]. To further investigate corneal stromal remodeling, the protein expression levels of matrix metalloproteinases 3 (MMP-3) and 9 (MMP-9) were analyzed by Western blot. As shown in Fig. 7M, the BAK-induced DED model mice exhibited upregulated expression of both MMP-3 and MMP-9, indicating accelerated ECM degradation. Treatment with CELNs alone slightly reduced MMP expression, suggesting their inherent matrix-protective property. In the CsA-treated group, MMP-9 expression was significantly suppressed, whereas the decrease in MMP-3 was less pronounced, a difference that may be attributed to the indirect modulation of inflammatory pathways by CsA [51]. Notably, CsA@CELNs treatment resulted in the most substantial downregulation of both MMP-3 and MMP-9, with protein levels approaching those in the normal control group. These results demonstrate the superior efficacy of CsA@CELNs in maintaining corneal stromal integrity by effectively inhibiting key MMPs.

Safety evaluation

In vitro hemolysis tests were first conducted to assess the safety of CsA@CELNs eye drop. Both CsA@CELNs and formulated CsA@CELNs eye drop caused minimal erythrocyte damage at the set concentrations, with hemolysis rates consistently below 5% (Fig. 8A-D), demonstrating excellent blood compatibility.

Fig. 8.

Safety evaluation of CsA@CELNs. (A) Quantitative hemolysis ratio of CsA@CELNs in 2% erythrocyte suspensions. (B) Representative photographs of CsA@CELNs in distilled water, 0.9% NaCl solution, and 2% erythrocyte suspensions. (C) Quantitative hemolysis ratio of CsA@CELNs eye drop in 2% erythrocyte suspensions. (D) Representative photographs of CsA@CELNs eye drop in distilled water, 0.9% NaCl solution, and 2% erythrocyte suspensions. (E) Ocular surface photographs of rats. (F) Fluorescein sodium staining images of the ocular surface. (G) Representative H&E sections of the cornea (scale bar: 50 μm). (H) Number of blinks per minute after administration of different eye drops. Wetted length of the sodium fluorescein test strip immediately (I) and 10 min (J) after instillation of eye drops. (K) Representative H&E sections of major organs (scale bar: 100 μm). All the quantitative data are means ± SD; n = 3. (*P < 0.05, ****P < 0.0001)

Previous studies have indicated that CsA eye drops may cause some degree of irritation to the ocular surface [52, 53]. To improve their safety profile, this study utilized biocompatible CELNs to encapsulate CsA, aiming to reduce its potential irritancy. For the ocular irritation assessment, healthy rats were selected as the experimental animals to facilitate accurate observation of local reactions such as conjunctival hyperemia, corneal edema, and secretions. After administering PBS, lkervis^®, and CsA@CELNs eye drop twice daily (8 am and 4 pm) for 5 consecutive days, the results showed that compared to the normal group, the CsA group exhibited a slight increase in intraocular blood vessels (Fig. 8E). However, no sodium fluorescein staining was observed on the ocular surface (Fig. 8F), indicating that the integrity of the tear film remained intact. H&E staining also revealed no significant structural changes in the cornea was observed (Fig. 8G). All assessed indicators in the CsA@CELNs group were comparable to those in the normal controls, demonstrating the formulation’s favorable safety profile and absence of irritating effects.

Serving as a primary defensive reflex against noxious stimuli, blinking frequency acts as a direct indicator of ocular pain and discomfort levels [54]. We administered 10 µL of PBS, Ikervis^® and CsA@CELNs eye drop and counted the blinks within 1 min. Rats in the Ikervis^® group exhibited rapid and frequent blinking after administration, whereas the PBS and CsA@CELNs groups showed significantly lower blink frequencies (Fig. 8H). Moreover, tear secretion was also measured as it serves as a standard indicator of reflex irritation [55]. The wetting length of Schirmer’s test strips in the Ikervis^® group was markedly greater than in the other two groups after administration, indicating that raw CsA can elicit an acute irritant response (Fig. 8I). After 10 min, tear production showed no significant differences among the three groups (Fig. 8J). These findings demonstrated that while CsA induces transient ocular irritation, this effect resolves spontaneously within minutes. Encapsulation of CsA in CELNs significantly attenuates this irritation, thereby improving the safety profile of its administration.

Although CsA@CELNs were administered topically to the eye, they may still be absorbed through the nasolacrimal duct or penetrate the ocular surface into systemic circulation, posing a potential risk of toxicity to distant organs [56]. To address this, histopathological analysis was performed on major organs such as the heart, liver, spleen, lungs, and kidneys. H&E staining (Fig. 8K) showed that the tissue structures of all examined organs under CsA@CELNs treatment were clear, with no significant pathological changes compared to the normal group, indicating that the formulation did not cause systemic toxicity. This further supports the safety profile of CsA@CELNs eye drop for clinical application.

Conclusion

In conclusion, we have successfully engineered a novel eye drop system, CsA@CELNs, by leveraging biocompatible CELNs as a natural nanocarrier for CsA to treat DED. This system features a uniform particle size, high drug loading, and favorable release properties, which collectively contribute to significantly enhanced corneal permeability and prolonged ocular surface retention. Functionally, CsA@CELNs operate through a synergistic multi-target mechanism to disrupt the “oxidative stress-inflammation-tissue damage” vicious cycle of DED pathogenesis. CsA@CELNs efficiently scavenges excess ROS and restores mitochondrial membrane potential, thereby markedly alleviating oxidative stress via potent activation of the Nrf2/HO-1/NQO1 signaling pathway. Concurrently, the formulation suppresses the NF-κB pathway, reduces the secretion of key pro-inflammatory cytokines (TNF-α, IL-1β, IL-6), and promotes macrophage repolarization from a pro-inflammatory M1 to an anti-inflammatory M2 phenotype. In a BAK-induced mouse DED model, CsA@CELNs eye drop effectively restored tear secretion, promoted the regeneration of corneal epithelial and conjunctival goblet cells, and improved lacrimal gland histology, thereby comprehensively alleviating ocular surface damage.

Recent advancements in ocular drug delivery have highlighted the versatility of EV-based strategies. Ren et al. [57] utilized milk-derived exosomes combined with ectoine to optimize storage stability. Xie et al. [58] developed a DNA-zipper-mediated membrane fusion technique to functionalize corneal exosomes for targeted gene silencing. In our study, the CsA@CELNs eye drop leverage PDEVs as a natural delivery vehicle. These plant vesicles inherently contain antioxidant and anti-inflammatory components that function synergistically with the encapsulated CsA, obviating the need for synthetic excipients and enhancing overall biocompatibility. Although plant derived EVs have not previously been investigated for DED, our findings demonstrate that this system not only constitutes a promising therapeutic option but also establishes a versatile platform, significantly expanding the potential of plant-derived EVs for ocular drug delivery and broader their biomedical applications.

Supplementary Information

Below is the link to the electronic supplementary material.

Author contributions

J. Li and P. Lee designed the research work, performed the experiments, analyzed the data, and wrote the original manuscript. C.Shi, D. Mao, X. Liu and Z. Yan performed the experiments and analyzed the data. H. Zou and M. Cai analyzed and curated the data. Z. Wu revised the manuscript. Y. Lyu and N. Feng acquired the funding, supervised the experiments and revised the manuscript.

Funding

This study was supported by the National Natural Science Foundation of China (No. 82505359), Shanghai Sailing Program (23YF1447300, China) and the Program for Shanghai High-Level Local University Innovation Team (SZY20220315, China).

Data availability

No datasets were generated or analyzed during the current study.

Declarations

Ethics approval and consent to participate

All animal procedures followed the guidelines set by the Regional Ethics Committee for Animal Experiments and adhered to the Care Regulations approved by the Institutional Animal Care and Use Committee at Shanghai University of Traditional Chinese Medicine (PZSHUTCM2505090020, PZSHUTCM2510170002).

Consent for publication

All the authors agree to the publication of the article.

Competing interests

The authors declare no competing interests.