THO initiates inflammatory signaling cascades by activating ion channel-mediated pathways in both corneal epithelial cells and infiltrating immune cells within the ocular surface. This results in activation of transcription factors that upregulate genes encoding inflammatory cytokines, ultimately leading to robust inflammatory responses in DED patients. Persistent uncontrolled immune activation disrupts the structure and function of ion channel proteins, thereby reinforcing THO and worsening symptoms^[45]. Corneal epithelial cells and sensory neurons respond to hyperosmolarity at thresholds of approximately 600-700 mOsm/kg. Neurons mediate nociceptive signaling, whereas epithelial cells activate inflammatory signaling^[41].

THO induces conformational changes in TRPV1 channels, allowing excessive cation influx and disrupting cellular homeostasis^[45,64]. Increased intracellular calcium, often partly via TRPV1, subsequently activates transcription factors such as MAPKs, activator protein 1 (AP-1), and NF-κB. These events trigger the secretion of pro-inflammatory cytokines, chemokines, and matrix metalloproteinases (MMPs), escalating inflammatory responses in a cascade-like manner^{[1,19,48,49,51,65-68]}. MAPKs constitute a family of highly conserved signaling pathways consisting of three major members: p38 MAPK, extracellular signal-regulated kinase (ERK), and JNK. By activating MAPK pathways, hyperosmolarity induces a pro-inflammatory response in both human limbal epithelial cells and corneal epithelial cells via IL-1β, tumor necrosis factor-α (TNF-α), MMP-9, and the C-X-C chemokine IL-8^{[19,51,69-71]}. Elevated IL-1β and TNF-α further drive MAPK activation and promote MMP-9 expression^[71]. In DED, inflammatory cascades are promoted by MMPs, which act through cleavage of pro-cytokines and additional extracellular proteins (e.g., receptors, growth factors, adhesion molecules) and through the establishment of chemokine gradients^[72]. The levels of MMP-9, one of the most important MMPs on the ocular surface^[71], correlate positively with tear osmolarity, for which it serves as a potential biomarker^[65]. Additionally, Toll-like receptors (TLRs) on epithelial cells activate NF-κB via the myeloid differentiation primary response protein 88 (MyD88)-dependent pathway, further amplifying inflammatory gene expression^[73,74]. Figure 3 summarizes the initiation and signaling pathways of inflammation in DED.

Hyperosmotic and desiccating stress induces ROS in human corneal epithelial cells, promoting IL-1β production via the downstream NLRP3/caspase-1/IL-1β axis^[54]. Additionally, TLR4 stimulation by desiccating stress could activate the NLRP12 and NLRC4 inflammasomes via caspase-8 signaling^[56]. Inflammasomes serve as critical elements of innate immunity, facilitating the maturation and secretion of IL-1β and IL-33^{[1,51,55,56,58,75-77]}. IL-1β bridges innate and adaptive immunity, serving as a pro-inflammatory hallmark of DED onset and progression^[75]. Similarly, IL-33, an important member of the IL-1 family, serves as a pro-inflammatory cytokine that mediates ocular surface alterations in DED^[56].

IL-1β and TNF-α stimulate multiple kinases and transcription factors within dendritic cells (DCs) infiltrating the eye, thereby driving DC-mediated naïve T cell differentiation into Th1 and Th17 subsets^[68]. A cascade of kinase activity leads to phosphorylation and activation of the transcriptional regulators c-Jun and activating transcription factor 2 (ATF2). Following activation, these two regulators migrate to the nucleus and upregulate the production of pro-Th1 (IL-12) and pro-Th17 (IL-1β, IL-6, IL-23) cytokines in DCs. IL-12 released by DCs activates the T-box protein expressed in T cells (T-bet) as well as STAT1 in naïve CD4⁺ T lymphocytes, consequently steering them toward an IFN-γ-producing Th1 phenotype. Likewise, IL-1β, IL-6, and IL-23 released by DCs activate STAT3 and RAR-related orphan receptor gamma T (RORγT) in naïve CD4⁺ T cells, fostering the growth and maturation of inflammatory CD4⁺ Th17 cells capable of producing IL-17 and IL-22^[19,68,78].

Th1-derived IFN-γ together with Th17-derived IL-17 and IL-22 further stimulates macrophages and neutrophils on the ocular surface in DED patients. This induces M1 macrophage polarization and release of pro-inflammatory cytokines, chemokines, MMPs, adhesion molecules, and pro-lymphangiogenic molecules, followed by recruitment of monocytes and neutrophils to the inflammatory site^[72,79]. Th1 C-X-C motif chemokine ligands CXCL9, CXCL10, and CXCL11 and Th17 C-C motif chemokine ligand CCL20, released by M1 macrophages and neutrophils, recruit and anchor circulating Th1 and Th17 cells to the ocular inflammation sites. This process subsequently results in substantial infiltration of CD4⁺ Th1/Th17 cells into the inflamed ocular surface in DED patients^[79-81]. Under physiological conditions, regulatory T cells (Tregs) help maintain immune homeostasis and promote peripheral tolerance^[78,82]; however, an imbalanced Th17/Treg ratio disrupts this equilibrium and contributes to disease progression^[73]. The resulting Th1- and Th17-mediated inflammation drives tissue damage through distinct but complementary mechanisms. Th1-type inflammation promotes apoptosis of ocular epithelial cells, squamous metaplasia, and goblet cell loss, whereas Th17 cytokines damage lacrimal and meibomian glands and reduce mucin production. In parallel, activation of MMPs contributes to disruption of the corneal epithelial cell barrier^{[46,72,80,83]}. The ocular surface pathology in DED promoted by M1 macrophages and neutrophils is characterized by diminished mucin secretion and the combined loss of corneal epithelial cells, conjunctival epithelial cells, and goblet cells through apoptosis^[79].

Repeated desiccating stress also induces epithelial-mesenchymal transition, generating fibroblast-like epithelial cells enriched in IL-2/STAT5 signaling, IL-6/JAK/STAT3 signaling, and the inflammatory response pathway, with increased levels of multiple chemokines, including CXCL1, CXCL10, CCL2, and CCL19. These findings suggest that fibroblast-like epithelial cells primarily recruit immune cells and promote acute inflammation in DED. The crosstalk between these cells and macrophages/CD4⁺ T lymphocytes constitutes a critical basis for the subsequent inflammatory cascades in DED^[84].

In summary, the pathogenesis of DED involves the dual activation of innate and adaptive immunity^[85]. Under THO and desiccating stress, the damaged corneal epithelium releases inflammatory cytokines that promote the maturation of resident antigen-presenting cells. These cells subsequently capture antigens and migrate to draining lymph nodes, where they activate naïve T cells that then home to the ocular surface^{[46,81,85,86]}. Infiltrating T cells cause further tissue injury and reactivate innate inflammation, creating a self-sustaining cycle of “inflammation → damage → re-inflammation” that drives DED onset and persistence^[85,87].

MSC-EVs exert potent anti-inflammatory and immunoregulatory effects by transferring bioactive molecules, including regulatory miRNAs and immunomodulatory proteins, into infiltrating immune cells at the ocular surface^{[102,123-125]}. miRNAs are small non-coding RNA molecules, typically 20-22 nucleotides in length, that modulate gene expression by influencing the stability of mRNAs and interfering with protein biosynthesis, consequently altering the phenotype and function of immune cells^[126].

Suppression of inflammation via targeting of core inflammatory pathwaysA key target of MSC-EVs in DED therapy is the NF-κB pathway. This pathway is a master regulator of inflammation, immune response, and cell survival. The canonical NF-κB signaling cascade involves key components such as TLR4, IL-1 receptor-associated kinase 1 (IRAK1), TNF receptor-associated factor 6 (TRAF6), transforming growth factor beta 1 (TGF-β) activated kinase 1/MAP3K7 binding protein 2 (TAB2), and NF-κB inhibitor epsilon (NFKBIE), among others^[127].

Several studies show that human UCMSC-EVs are enriched with miRNAs such as miR-125b, let-7b, and miR-6873, that hit multiple upstream parts of the NF-κB pathway. In a DED mouse model, these miRNAs target TLR4, IRAK1, TAB2, and TRAF6, reversing the pathway’s abnormal activation. This multi-pronged inhibition sharply lowers pro-inflammatory mediators (like TNF-α, IL-4, IL-8, IL-10, and IL-17) while boosting protective cytokines such as IL-13 in the cornea and conjunctiva. Tear cytokine profiling further confirmed a broad drop in inflammatory cytokines, including IL-2, IL-6, IL-7, IL-11, IL-12p70, IL-15, IL-17, IL-1Ra, IL-10, TNF-α, IFN-γ, IL-5, and the chemokines CCL2, CXCL1, and IL-8. Conversely, the tissue inhibitors of MMPs, TIMP-1 and TIMP-2, were markedly elevated. Collectively, these findings indicate that topical administration of human UCMSC-EVs alleviates DED symptoms, suppresses ocular inflammation, and restores corneal epithelial homeostasis through multi-targeted modulation of the IRAK1/TAB2/NF-κB pathway^[127]. Likewise, bone marrow mesenchymal stem cell-derived exosomes (BMSC-Exos) deliver miR-21-5p to inhibit the TLR4/MyD88/NF-κB pathway. BMSC-Exo treatment significantly lowers tear and peripheral blood IL-17 and IL-22 concentrations while raising those of IL-4, IL-10, and TGF-β1, ultimately alleviating symptoms in both DED patients and mouse models^[73].

In Sjögren’s syndrome-associated DED models, lacrimal gland injection of MSC-EVs significantly downregulated the mRNA levels of pro-inflammatory mediators (TNF-α, IL-1β, and IFN-γ) in both the ocular surface and lacrimal glands while mitigating T-cell infiltration within the lacrimal glands^[106]. Mechanistically, MSC-EVs inhibited the downstream TLR4 and T cell receptor signaling pathways in activated mouse splenocytes, thereby inhibiting the nuclear translocation of nuclear factor of activated T-cells 1 (NFAT1) and NF-κB. In TLR4-stimulated splenocytes, MSC-EVs also inhibited the nuclear translocation of NF-κB and p38. Notably, let-7b-5p encapsulated in the MSC-EVs was identified as a negative regulator of the TLR4/NF-κB pathway^[106]. Levels of the downstream effector MMP-9 can increase through MAPK signaling activation by ocular epithelial and immune-cell-derived pro-inflammatory cytokines such as IL-6^[128,129]. Clinically, topical ophthalmic administration of human Wharton’s jelly-derived MSC exosomes (WJMSC-Exos) reduced the gene expression of pro-inflammatory cytokines (such as IL-6 and IL-22) as well as MMP-9 in patients with Sjögren’s syndrome-associated DED^[128]. Tear MMP-9 levels dropped one month after treatment, suggesting suppression of extracellular matrix (ECM) degradation and inflammation.

Additionally, miR-223-3p delivered in mouse adipose-derived MSC exosomes (ADMSC-Exos) has demonstrated potent anti-inflammatory effects in DED models through targeting F-box and WD repeat domain-containing 7 (Fbxw7) and subsequent significant reduction of pro-inflammatory cytokines (IL-6, IL-17, TNF-α, and IL-1β) and chemokines (CCL2 and CXCL1)^[130]. Based on existing evidence that deficiency of the ubiquitination of enhancer of zeste homolog 2 (EZH2) promotes TNF-α-triggered NF-κB activation in intestinal epithelial cells, it was hypothesized that miR-223-3p may block EZH2 degradation by targeting the suppression of Fbxw7 expression, consequently inhibiting the inflammatory response in a DED model. Moreover, miR-223 has been shown to directly suppress NLRP3 inflammasome activity by inhibiting NLRP3 translation, thereby suppressing the inflammatory response in DED^[76]. Several studies further confirm that adipose-derived MSC-derived EVs (ADMSC-EVs) can inhibit the assembly of desiccation stress-induced NLRP3 inflammasomes, the activation of caspase-1, and the maturation of IL-1β^[131-133]. By suppressing the NLRP3/IL-1β signaling axis, inflammatory cytokines (e.g., IL-1β, IL-6, IL-1α, IFN-γ, and TNF-α) were reduced while the anti-inflammatory cytokine IL-10 was elevated, effectively mitigating DED-related ocular surface damage.

Immunomodulatory effects of MSC-EVs in the treatment of DEDMSC-EVs can suppress DC-mediated inflammation, thereby alleviating ocular surface immune responses in DED. MSC-EVs have a general inhibitory effect on DCs. They inhibit the antigen uptake capacity of immature DCs, alter their morphology, reduce their dendrite number, and suppress both the increase in DC population and their maturation in DED. Consequently, the cytokine production profile of DCs shifts from an inflammatory to an immunoregulatory phenotype. MSC-EVs downregulate pro-inflammatory cytokines such as IL-6 and IL-12p70 and upregulate the anti-inflammatory cytokine TGF-β, thereby inhibiting DC proliferation and migration^[14,134]. Moreover, miR-21-5p enriched in MSC-EVs may contribute to this inhibitory effect on DC function by downregulating CCR7 expression and impairing the chemotactic response to CCL21^[14]. Collectively, these effects reduce Th17 cell activation, a key driver of ocular surface inflammation in DED.

Macrophages are critical innate immune cells with high plasticity and play dual roles in inflammation and tissue repair. Their specific functions are determined by their polarization status, which is shaped by signals from the local microenvironment. Conventionally, macrophages are divided into a pro-inflammatory M1 (classically activated) and an anti-inflammatory M2 (alternatively activated) phenotype^[135-137]. MSC-EVs can suppress the activation of pro-inflammatory M1 macrophages while driving their polarization towards the anti-inflammatory M2 phenotype^[21,135]. Human UCMSC small EVs (UCMSC-sEVs) mitigate autoimmune-mediated dacryoadenitis, possibly through delivering encapsulated miR-100-5p. This miRNA induces M2-polarized macrophages and enhances Treg populations in vivo and in vitro, thereby reducing inflammation and facilitating tissue repair. Following human UCMSC-sEV treatment in rabbit lacrimal glands, the mRNA levels of M1-associated genes - nitric oxide synthase 2 (NOS2), interferon-regulatory factor 5 (IRF5), TNF-α, IL-1β, and IL-6 - were markedly reduced, whereas those of M2-related genes such as arginase-1 (Arg1), mannose receptor (CD206), Krueppel-like factor 4 (KLF4), IL-10, and TGF-β were significantly upregulated^[135]. M1 macrophages produce high levels of inflammatory mediators, including TNF-α, IL-1β, IL-6, IL-12, and IL-23. These cytokines collectively serve as master regulators of inflammatory responses, while also recruiting and activating additional immune cells and ultimately contributing to tissue damage. In contrast, M2 macrophages predominantly produce anti-inflammatory cytokines such as IL-10 and TGF-β, which suppress inflammation and promote tissue repair^[136]. In patients with graft-versus-host disease-associated DED refractory to currently available drugs or surgeries, eye drops containing exosomes from mesenchymal stromal cells (MSC-Exo) suppressed the infiltration and activation of ocular surface macrophages. MSC-Exo effectively regulated the phenotypic switching of macrophages, suppressed the expression of pro-inflammatory genes (e.g., IL-6, IL-1β, IL-17A, CD86), reestablished immune balance on the ocular surface, and consequently alleviated inflammatory damage. Mechanistically, MSC-Exo containing miR-204 directly engages interleukin-6 receptor (IL-6R), mediating the downregulation of IL-6/IL-6R/STAT3 signaling, which then induces the phenotypic switching of pro-inflammatory M1 macrophages to anti-inflammatory M2 macrophages on the ocular surface and thus alleviates dry eye symptoms^[21]. The uptake of exosomes derived from periodontal ligament mesenchymal stem cell (PDLSC-Exos) by inflammatory M1 macrophages promotes their polarization toward the immunosuppressive M2 phenotype, characterized by elevated expression of IL-10 and Arg1. This polarization shift partially contributes to the protective effect of PDLSC-Exos on conjunctival goblet cells against M1 macrophage-mediated inflammation^[137].

MSC-EVs can reduce neutrophil infiltration in the corneas of rats with DED. Specifically, umbilical cord MSC-derived exosomes (UCMSC-Exos) disrupt neutrophil recruitment through multiple mechanisms, including suppression of chemotaxis, reduction of neutrophil extracellular trap (NET) formation, and modulation of apoptosis^[129].

MSC-EVs can significantly suppress CD4⁺ T cell proliferation and regulate their differentiation. Tregs, which characteristically express the forkhead transcription factor Foxp3, possess immunosuppressive functions. By increasing Treg populations while inhibiting Th17 cells, the Th17/Treg balance is effectively restored, ultimately leading to a marked immunosuppressive effect^[135]. For instance, subconjunctival injection of human UCMSC-sEVs promoted an M1-to-M2 macrophage phenotypic shift in a rabbit model of dry eye, thereby suppressing the proliferation of CD4⁺ T cells and increasing the proportion of CD4⁺ Foxp3+ Tregs. MiR-100-5p has been identified as being at least partially responsible for the T cell suppression and Treg expansion mediated by macrophages incubated with human UCMSC-sEVs^[135]. Further supporting their role in T cell modulation, human UCMSC-EVs were shown to inhibit the infiltration of CD4⁺ T cells into the conjunctiva of DED mice^[127]. Similarly, BMSC-Exos alleviate DED progression through restoring the Th17/Treg balance by inhibition of the TLR4/MyD88/NF-κB pathway in CD4⁺ T cells. BMSC-Exo treatment significantly reduced levels of Th17 marker cytokines (IL-17 and IL-22) while increasing those of Treg-related cytokines (IL-4, IL-10, TGF-β1) in both tear fluid and peripheral blood of DED mice. Concurrently, the population of Th17-positive cells diminished and that of Treg-positive cells expanded, thereby restoring the Th17/Treg ratio^[73]. Consistent findings were observed in vitro: BMSC-Exos co-cultured with CD4⁺ T cells from DED patients downregulated IL-17 and IL-22 expression while upregulating IL-4, IL-10, and TGF-β1, leading to a reduced Th17 cell proportion and an expanded Treg population^[73]. The therapeutic potential of MSC-EVs extends to Sjögren’s syndrome, a chronic inflammatory autoimmune disease characterized by xerostomia and keratoconjunctivitis sicca (dry eyes)^[138]. In mice with experimental Sjögren’s syndrome, intraorbital injection of MSC-EVs attenuated the infiltration of T cells in the lacrimal glands. A strong positive correlation was found between the levels of key mediators TGF-β1, pentraxin 3 (PTX3), let-7b-5p, and miR-21-5p in MSC-EVs and their immunosuppressive function^[106]. Furthermore, in both non-obese diabetic mouse models and in vitro cultures of peripheral blood mononuclear cells from Sjögren’s syndrome patients, labial gland-derived MSC exosomes (LGMSC-Exos) suppressed Th17 cell differentiation and promoted Treg induction, thereby restoring the Th17/Treg balance. This immunomodulatory effect was accompanied by reduced production of IL-17, IFN-γ, and IL-6, alongside increased production of TGF-β and IL-10 by T cells, indicating partial suppression of inflammatory responses^[139]. In patients with primary Sjögren’s syndrome, human UCMSC-Exos suppressed excessive proliferation and apoptosis of CD4⁺ T cells and blocked cell cycle progression in the G0/G1 phase, thereby inhibiting S phase entry. Furthermore, UCMSC-Exos rebalanced the Th17/Treg ratio by reducing the proportion of Th17 cells and elevating that of Tregs, accompanied by a reduction in the levels of IFN-γ, TNF-α, IL-6, IL-17A, and IL-17F and elevation in the levels of IL-10 and TGF-β. Mechanistically, UCMSC-Exos regulate CD4⁺ T cell immunity in primary Sjögren’s syndrome by reducing the elevated autophagic flux in peripheral blood CD4⁺ T cells^[140]. Topical treatment with MSC-EVs in DED mice considerably decreased the Th17 cell population in the draining lymph nodes by inhibiting DC activation-mediated Th17 immune responses^[134].

MSC-EVs can also suppress B lymphocyte-mediated immune responses. When peripheral blood mononuclear cells from primary Sjögren’s syndrome patients were co-cultured with LGMSC-Exos, the proportion of CD19+CD20-CD24+CD38+ plasma cells significantly decreased. The underlying mechanism involves exosomal miR-125b from LGMSC-Exos binding to PR domain zinc finger protein 1 (PRDM1) mRNA and suppressing its translation, thereby inhibiting plasma cell differentiation. The miR-125b contained within LGMSC-Exos is essential for modulating autoantibody production and suppressing B cell-mediated autoimmune responses in primary Sjögren’s syndrome^[141]. PRDM1 (also designated B lymphocyte-induced maturation protein-1, BLIMP-1) is a transcription factor indispensable for directing B cell differentiation into plasma cells capable of antibody production.

Beyond their anti-inflammatory and immunomodulatory functions, MSC-EVs can exert reparative and regenerative functions by inhibiting apoptosis, promoting cell proliferation, and regulating angiogenesis through the bioactive molecules they carry^{[102,123-125,129,135,136]}. Extensive studies^{[21,73,106,127-132,134,135,137]} across murine, rabbit, and human models of DED and Sjögren’s syndrome-related ocular damage have demonstrated the comprehensive therapeutic benefits of MSC-EV treatment. MSC-EVs can restore ocular surface homeostasis, thereby providing protection against desiccating and hyperosmotic stress. Key improvements include enhanced tear film stability (evidenced by prolonged tear film breakup time and increased tear secretion), reduced ocular surface disease index (OSDI) scores, and significant alleviation of corneal epithelial damage (evidenced by fluorescein staining) and conjunctival hyperemia. Moreover, substantial mitigation of corneal damage, together with improved epithelial integrity have been consistently observed. These effects are characterized by a well-organized arrangement of corneal epithelial cells, stromal cells, nerves, and endothelial cells, as well as increased numbers of corneal epithelial layers, thickening of the corneal and conjunctival epithelia, and active proliferation of corneal stromal cells. Concurrently, MSC-EVs effectively suppressed apoptosis in both corneal stromal cells and conjunctival epithelial cells, while promoting conjunctival goblet cell proliferation and upregulating the expression of their specific proteins, mucin-1 (MUC-1) and mucin-5AC (MUC-5AC). By specifically targeting myofibroblast activation and differentiation, UCMSC-Exos exhibit anti-fibrotic potential. The observed reduction in myofibroblast formation indicates their potential to prevent corneal fibrotic changes that lead to scarring and impaired vision^[129].

MSC-EVs promote corneal wound healing and repair through the action of their diverse cargo of growth factors. For example, ADMSC-Exos enhance the growth and plasticity of corneal stromal cells by inhibiting MMPs, upregulating ECM-related proteins including collagens and fibronectin, and thereby promoting ECM synthesis^[142]. The inhibition of MMP activation effectively prevents collagen degradation and matrix thinning, thereby indirectly maintaining matrix homeostasis^[129,142]. Furthermore, in mouse models of DED, treatment with human UCMSC-EVs altered the cytokine profile in tears, elevating epidermal growth factor (EGF) concentrations while reducing vascular endothelial growth factor (VEGF) levels. Notably, in a rat DED model, administration of UCMSC-Exos significantly reduced the length and branching of corneal neovessels. These consistent anti-angiogenic effects across species suggest the presence of specific anti-angiogenic factors or regulatory molecules within human UCMSC-EVs may contain anti-angiogenic factors r regulatory molecules capable of modulating genes involved in angiogenesis. Thus, human UCMSC-EVs demonstrate potential to simultaneously promote corneal regeneration and suppress pathological neovascularization^[127,129]. Treatment of Sjögren’s syndrome-associated DED with topical ophthalmic human WJMSC-Exos led to the upregulation of genes encoding multifunctional healing-promoting peptides, including lactoferrin, EGF, and thrombospondin-1 (TSP-1)^[128]. EGF, a potent polypeptide mitogen, is present in both the tear film and serum. Synthesized and released by the lacrimal gland, it plays a key role in maintaining corneal epithelial integrity and facilitating wound healing^[128]. Lactoferrin constitutes approximately 25% of total tear proteins and alleviates dryness in DED patients by reducing oxidative stress and inflammation^[62]. Oral administration of enteric-coated lactoferrin capsules at a daily dose of 270 mg has also demonstrated efficacy, promoting goblet cell secretion and extending tear film break-up time^[143]. TSP-1 is a key activator of latent TGF-β, thereby enabling the immunomodulatory and wound-healing activities of TGF-β. In addition, TSP-1 is also a potent endogenous inhibitor of angiogenesis. Although this anti-angiogenic effect has not been explicitly demonstrated in the cornea, TSP-1 is known to exert its anti-angiogenic effects through other critical signaling pathways^[144].

Current research on the mechanisms of MSC-EVs in the treatment of DED largely focuses on their anti-inflammatory and immunomodulatory effects, whereas their effects on tissue repair and regeneration remain insufficiently explored. In contrast, these regenerative functions have been more widely studied in other non-DED corneal disorders. Treatment with human UCMSC-sEVs has been demonstrated to remarkably improve the proliferation and migration of corneal epithelial cells in vitro, as well as corneal epithelial wound healing in vivo, by delivering miR-21, which suppresses PTEN and consequently activates the PI3K/Akt signaling pathway^[145]. Similarly, EVs from bone marrow MSCs (BMMSC-EVs) promote corneal epithelial repair by enhancing cell proliferation and inhibiting apoptosis through the downregulation of pro-apoptotic genes encoding BCL-2-associated agonist of cell death (BAD), BCL-2-associated X-protein (BAX), P53, and caspase 3 and the concurrent upregulation of the anti-apoptotic gene encoding B-cell lymphoma 2 (BCL-2)^[146,147]. ADMSC-Exos facilitate corneal endothelial wound healing and promote the migration of corneal epithelial cells both in vitro and in vivo through a dose-dependent induction of the G0/G1-to-S-phase cell cycle shift, inhibition of ROS-mediated senescence, and suppression of autophagy - thereby inhibiting epithelial-mesenchymal transition while restoring mitochondrial function and promoting corneal epithelial regeneration^[148]. In addition, human BMMSC-EVs enhance regeneration of mouse trigeminal ganglion neurons. Notably, EVs derived from three-dimensional (3D) culture systems significantly promote neurite elongation, branching, and complexity compared with those derived from two-dimensional (2D) cultures^[122].

In summary, leveraging their anti-inflammatory, immunomodulatory, reparative, and regenerative properties, MSC-EVs demonstrate considerable therapeutic potential for DED at the molecular, cellular, histological, and functional levels [Figure 4].

Table 1 summarizes in vivo studies using MSC-EVs in dry eye and related diseases. The study by Habibi et al. reported a randomized controlled trial in patients with primary Sjögren’s syndrome that employed a contralateral-eye design and assessed tear protein biomarkers (e.g., IL-6, MMP-9, and EGF)^[128]. However, the small sample size (n = 8) and the lack of characterization of specific exosomal cargo limit the strength and generalizability of its conclusions. Mechanistic insights are strengthened by the study of Zhou et al., which employed both loss- and gain-of-function approaches to support a causal role for miR-204^[21]. In contrast, most other studies rely primarily on inhibitor-based or correlative analyses, resulting in weaker causal inference. From a methodological perspective, the study by Kim et al. represents a rigorous approach, systematically comparing early-passage (15 population doublings, PD15) vs. late-passage (PD40) MSCs, and 2D vs. 3D culture conditions, while integrating proteomics and miRNA sequencing to evaluate EV function^[106]. These findings provide a useful framework for EV production and quality control, although human validation remains lacking. Additionally, the study by Zhao et al. established a mechanistic-phenotypic association by implicating the miR-21-5p/TLR4/MyD88/NF-κB axis in restoring Treg/Th17 balance^[73], although this was demonstrated in a benzalkonium chloride-induced chemical injury model that may not fully recapitulate autoimmune dry eye pathology.

Many studies have utilized topical administration of MSC-EVs as eye drops^{[21,127-129,131,132]}, providing a non-invasive and convenient route that restores ocular surface immune homeostasis and mitigates inflammatory damage, highlighting strong translational potential.

Common limitations across studies include non-standardized dosing units, short follow-up (≤ 56 days in animals; ≤ 3 months in humans), and heterogeneity in MSC sources, isolation methods, and investigated miRNAs. Clinical validation remains limited, with only a few human studies, and most investigations remain preclinical, requiring further confirmation in well-designed clinical trials.

Current studies on engineered MSC-EVs in DED primarily focus on integration of functional cargo, particularly targeting the core pathological axis of “inflammation-ROS-injury”. Besides nanoparticle-based modification, cargo engineering can be achieved through techniques such as electroporation, sonication, and donor cell preconditioning to load small molecules, proteins, or nucleic acids^{[149,150,152,156,157]}.

A representative example of cargo engineering is MSCExo-Ce, a nanoparticle eye drop generated using in situ crystallization of cerium oxide nanocrystals onto MSC-derived exosomes^[154]. This system synergistically integrates the intrinsic anti-inflammatory and pro-regenerative properties of MSC-EVs with the strong ROS-scavenging capacity of cerium oxide. Experimental results demonstrated that MSCExo-Ce significantly promotes corneal epithelial cell migration and efficiently eliminates intracellular ROS in vitro. In a benzalkonium chloride-induced DED model, MSCExo-Ce markedly improved tear secretion and reduced corneal epithelial damage. It also suppressed IL-1β expression and inhibited apoptosis. Overall, it exhibited superior therapeutic efficacy compared with unmodified MSC-EVs, while maintaining excellent biocompatibility.

Another example is mExo@AA, developed by depositing ascorbic acid-reduced gold nanoparticles onto MSC-EV membranes^[155]. This engineered system preserves the native properties of EVs while adding better antioxidative and immunomodulatory functions. In vitro, mExo@AA accelerated epithelial wound healing and scavenged ROS. It also promoted M2 macrophage polarization and reduced pro-inflammatory cytokines like IL-6 and IL-1β. In vivo, it outperformed both EVs and ascorbic acid alone, showing synergistic effects on corneal repair, inflammation suppression, and tear restoration. Only the combined system completely cleared ROS - a clear demonstration of how cargo-engineered MSC-EVs can serve as multifunctional platforms.

Given the limited number of MSC-EV-focused studies, engineering strategies derived from non-MSC EV systems provide important complementary insights, particularly in improving targeting specificity and delivery efficiency. Hybrid engineering approaches, including membrane fusion, extrusion, and chemical conjugation, have also been explored to further enhance EV targeting and cargo delivery^{[149,150,157,158]}.

For example, a hybrid exosome vehicle was constructed by fusing small interfering RNA (siRNA)-loaded liposomes with corneal epithelial cell-derived EVs via DNA zipper-mediated membrane fusion^[158]. This design enables precise delivery of anti-NF-kappa-B inhibitor zeta gene (NFKBIZ) siRNA to corneal tissues, effectively silencing inflammatory signaling pathways. The hybrid exosome vehicle system reduced pro-inflammatory cytokine production and reprogrammed macrophages from an M1 to an M2 phenotype. It also decreased apoptosis and enhanced epithelial proliferation. In vivo, it significantly improved tear secretion and restored corneal integrity without detectable toxicity. Such hybrid engineering strategies demonstrate how EVs can be transformed from natural carriers into programmable delivery systems, offering valuable design principles that can be readily adapted to MSC-EV-based platforms.

Biomaterial-based delivery systems, particularly hydrogels, help engineered MSC-EVs overcome rapid clearance and limited ocular retention^[159,160]. Beyond hydrogels, alternative platforms such as nanoparticles, microneedles, and contact lens-based delivery systems have also been explored to improve ocular drug retention and controlled release^[161-163].

A thermosensitive chitosan-based hydrogel loaded with induced pluripotent stem cell-derived mesenchymal stem cells (iPSC-MSCs) enables sustained release and prolonged ocular surface residence^[159]. The system delivers miR-432-5p, which suppresses collagen production by targeting translocation-associated membrane protein 2 (TRAM2), thereby reducing fibrosis and promoting corneal epithelial and stromal repair. This approach improves corneal transparency while inhibiting neovascularization. Thermosensitive hydrogels effectively reduce drug loss caused by nasolacrimal drainage, significantly enhancing local bioavailability^[164].

Exos-miRNA-24-3p combined with a di(ethylene glycol) methyl ether methacrylate (DEGMA)-modified hyaluronic acid hydrogel enables sustained ocular delivery, accelerating epithelial wound closure, reducing stromal fibrosis, and suppressing macrophage activation^[160]. This system markedly improves corneal repair outcomes over extended treatment periods, demonstrating that hydrogel-assisted platforms can substantially enhance the clinical translation potential of engineered MSC-EVs by improving drug retention and delivery efficiency.

Beyond physicochemical and delivery-based modifications, genetic engineering of EV cargo is an emerging strategy for targeting immunoregulation. Approaches include miRNA enrichment, siRNA loading, and clustered regularly interspaced short palindromic repeats (CRISPR)-based modifications to precisely modulate disease-related pathways^{[149,150,157,158]}. EVs derived from LGMSCs and loaded with miRNA let-7f-5p showed significant therapeutic efficacy in a Sjögren’s syndrome model^[165]. Mechanistically, these engineered EVs targeted the 3’ untranslated region of RAR-related orphan receptor C (RORC), suppressing the RORC/IL-17A signaling axis and decreasing IL-17A production, which restored Th17/Treg immune balance. Therefore, engineered EVs can modulate disease-specific immune pathways at the molecular level, a precision medicine strategy for ocular surface inflammation.

Direct evidence for engineered MSC-EVs in DED remains limited. However, existing studies demonstrate that cargo modification, surface functionalization, hybrid systems, and genetic engineering can transform MSC-EVs into multifunctional, programmable platforms. The integration of insights from broader EV engineering strategies not only helps bridge current evidence gaps, but also highlights pathways for advancing MSC-EV-based therapies. As engineering technologies evolve, MSC-EVs are expected to transition from passive carriers to precisely controllable, mechanism-driven systems, thereby shifting DED management from symptom relief to true disease modification.

Although animal models of DED have yielded promising results, translating these findings is not straightforward due to major differences in anatomy, immunity, and physiology between animal and human eyes. For example, the human eyeball has a volume of about 6-7 mL, while the conjunctival sac, the main site where eye drops pool after topical application, holds only about 30 µL. In rodents (mice and rats), by contrast, the eyeball volume is merely 0.15-0.2 mL, and their conjunctival sac is proportionally even smaller^[151,166]. To circumvent these limitations, researchers in recent years have turned to 3D ocular surface models. Built from human cells, these in vitro systems recapitulate human ocular physiology faithfully at the cellular and microtissue level, thus helping to overcome the interspecies gaps^[167-170].

Recent work has shown that 3D ocular surface models can recreate the native structure of the cornea, lacrimal gland, and meibomian gland. As a result, these models offer more physiologically relevant platforms for studying DED than conventional 2D cultures or animal models. Consider long-term human corneal epithelial organoids derived from adult stem cells: they form a stratified epithelium, express functional markers like the ΔNp63α isoform of p63, keratin 12 (KRT12), and MUC1, and under hyperosmotic stress they upregulate pro-inflammatory cytokines (NF-κB1, IL-6, IL-8) while losing stemness, closely mimicking key aspects of DED pathophysiology^[170]. Human lacrimal gland organoids, generated from pluripotent stem cells or primary tissue, similarly keep their acinar and ductal features, like expression of aquaporin 5 (AQP5) and lysozyme (LYZ). They respond to cholinergic stimulation with calcium influx and LYZ release, and can serve as models for both Sjögren’s syndrome and age-related lacrimal gland hypofunction^[171-173]. For meibomian gland dysfunction, the leading cause of evaporative DED, both mouse and human meibomian gland organoids maintain holocrine lipid secretion. Studies using these organoids have also revealed that fibroblast growth factor (FGF) 10 is essential for their expansion and can prevent all-trans retinoic acid-induced meibomian gland atrophy in vivo^[174]. Meanwhile, microfluidic cornea-on-a-chip devices integrate an air-liquid interface, tear flow, and blinking mechanics, which allows real-time assessment of barrier function and drug permeability under shear-stress conditions relevant to DED^[168,175].

Researchers are increasingly using these 3D ex vivo models to screen drugs, investigate disease mechanisms, and develop regenerative therapies for DED. Corneal epithelial organoids have been successfully employed to evaluate clinically relevant DED drugs. For example, cyclosporine A suppressed inflammation, diquafosol improved cell viability, and basic FGF (bFGF) promoted proliferation, showing concordance with known clinical responses^[170]. Dome-shaped 3D corneal constructs, which have adjustable curvature, have allowed researchers to model ultraviolet B (UVB)-induced photokeratitis and to confirm the protective effect of eye drops containing bFGF^[176]. In addition, lacrimal gland and meibomian gland organoids are being explored for cell-based repair applications. Orthotopic transplantation of human meibomian organoids into immunodeficient mice resulted in engraftment, expression of mature meibocyte markers such as peroxisome proliferator-activated receptor gamma (PPARγ), and lipid production^[174]. Senescence-associated lacrimal gland organoids generated via magnetic 3D bioprinting showed that high mobility group box-1 (HMGB1) Box A gene therapy could prevent etoposide-induced apoptosis, restore ATP levels, and preserve secretory function^[177]. Altogether, these 3D ocular surface models bridge the gap between simple cell cultures and animal studies, offering predictive, human-relevant platforms for understanding DED pathogenesis and accelerating the development of novel therapeutics, including regenerative medicine approaches.

Importantly, the successes described above, including in vitro drug evaluations and in vivo animal transplantation studies, remain largely preclinical. Consequently, the potential of 3D ocular surface models has yet to bridge the gap from bench to bedside. Ma et al. observed that corneal organoid transplantation in humans has so far only been achieved using tissue fragments for ocular surface reconstruction. Transplanting a complete, intact corneal organoid is still only possible in animals^[178]. Similarly, current cornea-chip models still fall short of fully reproducing the cornea’s natural curvature, basement membrane, or innervation^[169]. Lacrimal gland organoids, for their part, continue to face hurdles in becoming vascularized, innervated, and functionally mature over the long term^[179]. Consequently, while these models hold promise for future clinical applications, robust human trials evaluating organoid- or chip-based therapies for DED have not yet been reported.

While organoid- and chip-based therapies remain preclinical, MSC-EVs have already entered clinical trials for DED. Inspired by the promising therapeutic effects of MSC-EVs observed in preclinical studies, as of June 2025, eight clinical trials investigating MSC-EVs for DED have been registered across ClinicalTrials.gov, ChiCTR, and the WHO ICTRP platforms [Table 2].

Obvious heterogeneity exists among these eight clinical trials regarding patient enrollment criteria, dosing regimens, cell sources, and reported outcomes.

(1) Heterogeneity in patient enrollment criteria. Age ranges vary from 20-60 years to 18-70 years. Disease subtypes include Sjögren’s syndrome, chronic graft-versus-host disease, post-refractive surgery, and non-specific dry eye. Objective severity thresholds, such as a Schirmer test score ≤ 10 mm/5 min, are employed as inclusion criteria in several studies.

(2) Heterogeneity in dosing and administration. Dosing strategies remain unstandardized. Three trials adopt a fixed concentration of 10 μg/drop, whereas one trial specifies a fixed volume of 0.125 mL/eye per dose. Administration frequency ranges from twice daily to four times daily. These discrepancies underscore the need for further dose-finding and regimen-optimization studies.

(3) Efficacy and results. To date, only two trials have published peer-reviewed efficacy data: one demonstrated statistically significant improvements in OSDI score, Schirmer I test, and tear break-up time (IRCT20211102052948N1)^[128]; the other reported clinical benefit in chronic graft-versus-host disease -associated dry eye, with mechanistic evidence implicating miR-204 cargo (NCT04213248)^[21].

(4) Cell source. Most trials utilize UCMSCs. One trial employs limbal stem cells, and another uses pluripotent stem cell-derived MSCs. The impact of differing cell sources on clinical outcomes remains to be elucidated.

Overall, although MSC-EVs show promising therapeutic potential for DED, particularly in two published trials, the field currently lacks standardized protocols and publicly available results from most registered studies. Standardization of manufacturing processes, dosing regimens, and outcome measures will be essential for future progress.

The first major translational challenge is overcoming the anatomical and physiological barriers inherent to the eye, particularly the rapid clearance of topically applied drugs by the tear film^[153,193]. For this reason, drug formulations, delivery devices, and administration routes need to be tailored to the specific therapeutic strategy to overcome these obstacles while ensuring adequate bioavailability. The goal is to increase ocular drug retention, slow down drug elimination, and ultimately improve treatment effectiveness^[193].

The tear film plays a critical role in maintaining ocular immune privilege and homeostasis, but it also limits the permeation and extended residence of conventional drug formulations, posing a significant challenge for topical ocular drug delivery. The mucus layer, aqueous layer, and corneal epithelium collectively influence the retention time, adsorption pattern, and penetration route of EVs on the ocular surface^[153,194]. Once a topical formulation reaches the precorneal pocket - the primary site for absorption and/or pharmacological action - it is rapidly cleared by ocular protective processes such as blinking, basal and reflex tearing, and drainage through the nasolacrimal system, resulting in low bioavailability and uneven distribution^[195,196]. Despite their small size and high biocompatibility, EVs are not exempt from these challenges.

Some advanced platforms, including hydrogels, nanoparticles, microneedles, and contact lens-based delivery systems, have shown promise in improving ocular retention and controlled release of EVs^[161-163].

Biomaterial-assisted EV delivery systems to prolong retentionEncapsulating EVs with mucoadhesive materials, such as thermosensitive chitosan or hyaluronic acid hydrogels, can enhance resistance to tear clearance, prolong ocular surface retention, and improve bioavailability^{[159,160,181]}. For example, the 3D-Exo-hydrogel system uses a photocurable gelatin methacryloyl (GelMA) hydrogel that rapidly gels upon visible-light irradiation, firmly adheres to the corneal stroma, and effectively fills defects. This system provides mechanical support while creating a stable seal that resists displacement. The porous structure of the hydrogel allows sustained release of 3D-cultured MSC exosomes (3D-Exos) over several days, maintaining therapeutic concentrations at the injury site without the need for repeated administration. Compared to 2D-Exos, 3D-Exos exhibit higher yield, enhanced anti-inflammatory and pro-proliferative effects, and superior capacity for tissue remodeling. By combining durable wound stabilization with prolonged exosome delivery, this cell-free system effectively reduces inflammation, promotes re-epithelialization, restores limbal stem cell function, minimizes scarring, and accelerates overall corneal healing^[197].

Surface engineering of EVs for enhanced penetration and retentionThe negative surface charge of EVs limits their penetration through the similarly negatively charged corneal stroma, which is rich in proteoglycans and collagens. Modifying exosomes with cationic motifs has been shown to significantly improve both corneal retention and penetration following topical application. For example, anchoring arginine-rich cationic peptides (CPC) or avidin via polyethylene glycol (PEG) 2000 lipid insertion neutralizes the native anionic surface charge of milk-derived exosomes from approximately -22 to -2 mV. The resulting weakly cationic surface enables reversible electrostatic binding to anionic glycosaminoglycans in the corneal stroma, reducing washout by the tear film and prolonging corneal retention. In addition, the cationic motifs induce a Donnan partitioning effect (K > 1 for Exo-CPC), increasing exosome accumulation at the corneal interface, while hydrophilic PEG chains facilitate diffusion through the stroma, resulting in a two-fold increase in steady-state diffusivity compared to unmodified exosomes^[198]. Similarly, incorporation of ε-polylysine-polyethylene glycol-distearoyl phosphatidylethanolamine (PPD) into EV membranes reverses their ζ-potential, enabling enhanced penetration and retention within negatively charged extracellular matrices. This strategy has significantly improved therapeutic efficacy in osteoarthritis and may be directly translatable to corneal applications^[199]. Overall, reversing EV surface charge toward a cationic state represents a promising approach to enhance both the penetration and retention of topically applied EV-based eye drops in the cornea^[181].

Targeting ligands, such as arginine-glycine-aspartic acid (RGD) peptides, can further enhance the selective uptake of EVs and their cargo by corneal cells through integrin-mediated binding, improving targeted delivery to corneal tissues^[181]. Dual-engineered nanoparticles, coated with MSC membrane vesicles and functionalized with Trans-Activator of Transcription (TAT) cell-penetrating peptides, demonstrate increased corneal permeability, strong adhesion to the ocular surface, and active homing to corneal neovascular endothelial cells. This combined functionality enables high local drug accumulation and sustained therapeutic effects^[200].

The second major challenge is systematically elucidating the pharmacological properties of MSC-EVs, such as bioavailability, targeting ability, pharmacokinetics, and biodistribution. A comprehensive understanding of these properties is essential for designing optimal administration regimens, including the appropriate dosage, delivery route, and treatment frequency to achieve precise and efficient targeting of ocular surface tissues^{[5,102,134,201-203]}. EVs can be administered through multiple routes, including intravenous, subconjunctival, subcutaneous, intraperitoneal, oral, intranasal, or intraocular delivery, with the choice depending on the therapeutic goal and the target tissue^[181,203]. Exosome absorption varies considerably depending on the route of administration, injection site, and dosage, all of which directly influence their therapeutic efficacy^[125,203].

Knowledge of the pharmacokinetic behavior of EVs in the chronically inflamed ocular surface microenvironment remains limited, particularly in the context of DED. Inflammatory alterations may significantly affect EV retention time, biodistribution, and biological activity. A robust, context-specific understanding of EV pharmacokinetics is therefore needed for the rational design of dosing regimens and treatment schedules. Addressing this knowledge gap will be critical for advancing EV-based therapies toward clinical translation and warrants greater focus in future studies.

The third major challenge is thoroughly establishing the long-term therapeutic profile of MSC-EVs. Their efficacy and safety need to be validated through comprehensive clinical trials, with particular emphasis on evaluating potential cumulative effects and unanticipated biological consequences^[134].

The fourth major challenge is establishing standardized Good Manufacturing Practice (GMP) production pipelines^{[5,104,153,192,203-210]}, including standardization of production processes (isolation, purification, extraction, and scalable production) and standardization of quality control (characterization, identification methods, protocols, and evaluation systems). Several key issues need to be addressed: (1) Control of heterogeneity and standardization of isolation: MSC-EVs exhibit inherent heterogeneity due to variations in their parental MSCs, cell passages, target cells, and the surrounding pathological environment^{[5,102,106,125,181,211]}. Consequently, exosomes obtained by different isolation methods vary in yield, purity, protein content, and membrane integrity. Whether these methodological differences impact the immunosuppressive function of MSC-Exos remains unclear^[125]. Currently, no unified gold standard exists for exosome isolation and purification, and commonly used techniques remain suboptimal with regard to sample purity, processing volume, and time efficiency^[102,209]. Therefore, standardized protocols for the collection, isolation, purification, and characterization of MSC-EVs should be established to fully elucidate their therapeutic potential, enable precise differentiation of EV subpopulations, and facilitate the development of standardized exosome potency assays. Such standardization is crucial to ensure comparability across studies, as well as to guarantee the quality, safety, and batch-to-batch consistency required for clinical applications; (2) Development of scalable manufacturing processes: Naturally produced exosomes generally exhibit low yields, and their isolation and purification efficiencies are often suboptimal, posing significant challenges for large-scale production^[102,192]. Moreover, current extraction processes frequently struggle to ensure consistent product purity and yield^[212]. To enable scalable EV production, it is essential to establish stable, cost-effective, and scalable production and purification processes. These efforts should be supported by advanced analytical techniques such as particle counting, size-based separation methods, and phenotypic characterization to ensure rigorous quality control^[115,213].

The fifth major challenge is establishing stable and reproducible storage conditions that preserve the therapeutic potential of EVs^{[102,203,211,214,215]}. The effects of different temperature ranges on the stability and bioactivity of EVs require systematic evaluation. Although numerous studies have examined the impact of storage temperature, freeze-thaw cycles, pH, and storage duration, the results remain inconsistent. Nevertheless, most studies suggest that storage at either 4, -20, or -80 °C all meet the requirements for EV preservation, with the optimal conditions depending on the intended application and storage duration. Although it has been reported that EVs remain relatively stable after multiple freeze-thaw cycles, minimizing such cycles is advisable to prevent degradation of bioactive cargo and loss of immunosuppressive effects^[125]. Storage duration should be determined according to the study design and temperature conditions: 4 °C is preferred for short-term storage, whereas -80 °C is recommended for long-term storage^[102,210]. A deeper understanding of how storage conditions affect EV structure, function, and biological activity is still needed to refine storage protocols and support the development of effective cryopreservation strategies.

The sixth major challenge is conducting rigorous toxicological and microbiological assessments to ensure the sterility of EV preparations. The potential presence of pathogens - such as bacteria, viruses, or fungi - in these products poses a significant risk of infection^[203,216]. In response to reported cases of sepsis, the International Society for Extracellular Vesicles has issued safety alert guidelines concerning unproven EV-based therapies^[217].

The final major challenge is establishing a globally unified regulatory framework to standardize the therapeutic application of EVs and ensure mandatory reporting of adverse events^[203]. In the United States and Europe, EVs are classified as drugs or biological products that require regulatory approval. Despite this, numerous businesses and clinics in these regions continue to promote unapproved exosome-based interventions for indications such as anti-aging, hair loss, autism, rheumatic diseases, and Parkinson’s disease. In contrast, in Japan EVs, including exosomes, are not legally defined as cellular products and therefore fall outside the scope of the Act on the Safety of Regenerative Medicine (ASRM)^[217]. These inconsistences highlight the urgent need for unified legislative standards, prohibition of unregulated profit-driven human applications to ensure patient safety.