Neuroinflammation and BBB dysfunction form an interdependent pathogenic module across AD, PD, ALS, and MS, with growing evidence supporting their involvement in HD. Chronic activation of microglia and astrocytes drives persistent release of inflammatory mediators such as tumour necrosis factor alpha (TNF‑α), interleukin (IL)‑1β, and IL‑6, leading to secondary damage of neurons, oligodendrocytes, and vascular cells^[16]. In parallel, endothelial cells, pericytes, and astrocytic endfeet downregulate tight-junction proteins such as claudin-5, occludin, and zonula occludens (ZO)-1, while upregulating matrix metalloproteinases (MMP-2 and MMP-9), leading to increased vascular permeability and immune-cell infiltration^[16]. These processes reinforce one another, establishing a self-amplifying loop in which persistent glial activation and BBB breakdown jointly drive immune dysregulation, parenchymal damage, and accelerated neurodegeneration.

Mounting evidence indicates that MSC-EVs may intervene within this cycle by reprogramming glial inflammatory states. In AD mouse models, treatment of MSC-EVs derived from umbilical cord (UC) and bone marrow (BM) reduced microgliosis and astrogliosis, decreased pro-inflammatory cytokine levels, and promoted a shift toward more homeostatic microglial phenotypes^[17,18]. In vitro, adipose tissue (AT)-derived MSC-EVs were efficiently internalised by microglia and attenuated Aβ-induced activation by suppressing nitric oxide production and inflammatory mediators^[19]. Comparable immunomodulatory effects are observed in ALS cellular models, where AT- and BM-MSC-EV treatment lowered nucleotide-binding oligomerisation domain (NOD)-like receptor family pyrin domain containing 3 (NLRP3) inflammasome activation and pro-inflammatory cytokines [TNF-α, IL-1β, IL-6, C-C motif chemokine ligand 2 (CCL2)], while restoring IL-10 expression in astrocytes derived from SOD1^G93A mice^[20,21]. Consistent with this mechanism, induced pluripotent stem cells (iPSC)-derived MSC-EVs attenuated neuroinflammation by targeting NLRP3 via EV-associated miR-223-3p in AD mouse models^[22]. Similarly, suppression of NLRP3 inflammasome activation has been shown to markedly protect dopaminergic neurons in the 1-methyl-4-phenyl-1,2,3,6-tetrahydropyridine (MPTP) mouse model of PD, highlighting the relevance of this pathway across neurodegenerative contexts^[23]. Evidence from human in vitro systems further supports this regulatory role. In interferon (IFN)-γ/TNF-α-stimulated human microglial cell lines, BM-MSC-EV exposure normalised microglial morphology and shifted inflammatory and metabolic proteomic profiles toward a less activated, more homeostatic state^[24]. Together, these findings support a model in which MSC-EVs may recalibrate the activation threshold of glial cells and constrain maladaptive inflammatory responses [Figure 1]^[25-27].

This functional reprogramming of glial inflammatory responses is underpinned, at least in part, by specific EV-derived regulatory RNAs that target conserved inflammatory signalling pathways. In experimental autoimmune encephalomyelitis (EAE) mice, BM-MSC-EVs naturally containing miR-181a-5p suppressed microglial inflammation and pyroptosis by targeting ubiquitin-specific protease 15 (USP15) and modulating the p65 subunit (RelA)/NIMA-related kinase 7 (NEK7) axes, resulting in reduced IL-1β and IL-18 production, decreased demyelination, and improved clinical scores^[28]. In ALS, IFN-γ-primed BM-MSC-EVs delivering miR-467f and miR-466q inhibited p38 mitogen-activated protein kinase (MAPK) signalling by targeting mitogen-activated protein kinase kinase kinase 8 (Map3k8) and Mk2, thereby reducing TNF-α and IL-1β secretion in both N9 microglia and primary microglia from SOD1^G93A mice^[29]. In an AD in vitro model, BM-MSC-EV-associated miR-146a attenuated astrocytic neuroinflammation by inhibiting nuclear factor kappa-light-chain-enhancer of activated B cells (NF-κB) signalling in response to Aβ exposure^[30]. Notably, hypoxia-conditioned MSC-EVs have been shown to dampen inflammatory signalling in AD mouse brains by reducing signal transducer and activator of transcription 3 (STAT3) phosphorylation and preventing nuclear translocation of NF-κB p65 in glial cells, demonstrating modulation of key inflammatory transcriptional regulators [Box 1]^[10,31-33]. These findings indicate that MSC-EV miRNAs converge on shared inflammatory pathways, including NF-κB, MAPK, and inflammasome pathways across disease contexts, enabling coordinated suppression of maladaptive glial activation [Figure 1].

Because sustained neuroinflammation directly compromises vascular integrity, effective immunomodulation is inseparable from stabilisation of the neurovascular unit. A growing body of evidence demonstrates that MSC-EVs exert direct protective effects on BBB structure and function. In EAE models, AT-MSC-EVs have been shown to inhibit C-X-C motif chemokine ligand 12 (CXCL12)-induced, integrin-dependent adhesion of T cells to endothelial cells via interactions with intercellular adhesion molecule (ICAM)-1 and vascular cell adhesion protein (VCAM)-1 in vitro^[34]. Consistent with this mechanism, MSC-EV treatment markedly reduced rolling and firm arrest of encephalitogenic T cells within spinal cord microvessels in vivo, thereby limiting immune-cell entry into inflamed CNS tissue^[34]. In context of CNS barrier dysfunction, UC-MSC-EV administration increased expression of tight-junction proteins such as claudin-5 and occludin, and reduced MMP-2 and MMP-9 expression^[27]. In vitro, BM-MSC-EVs also modulate human brain pericytes, altering morphology and cytokine secretion profiles in a manner that depends on the inflammatory priming state of the parent MSCs, further supporting a role for MSC-EVs in tuning neurovascular inflammatory signalling^[35].

BBB protection further extends to endothelial resilience under proteotoxic and inflammatory stress. In an α-synuclein aggregate-induced BBB disruption model relevant to PD, UC-MSC-EVs preserved endothelial barrier integrity, reduced TNF-α induction, and interfered with aggregate-membrane interactions^[36]. In vivo, neuroinflammatory regions in PD are characterised by elevated CCL2 expression, which promotes C-C motif chemokine receptor (CCR2)-dependent recruitment of peripheral immune cells to the substantia nigra. Leveraging this chemokine axis, MSC-EVs engineered to overexpress CCR2 preferentially accumulated in CCL2-rich inflamed regions in PD mouse models. This targeted localisation was accompanied by reduced peripheral immune-cell infiltration, together with attenuation of microglial activation and pro-inflammatory cytokine release^[37]. The anti-inflammatory effects of MSC-EVs are reinforced by their protein cargo, which includes vascular endothelial growth factor (VEGF), fibroblast growth factor (bFGF), and angiopoietin (Ang)-1, molecules known to regulate vascular stability and immune signalling^[38,39]. These proteins complement miRNA-mediated modulation of inflammatory and extracellular matrix pathways [Figure 1], suggesting that MSC-EV cargo may act through coordinated molecular programmes rather than isolated signalling events. Through this multi-layered cargo architecture, MSC-EVs can simultaneously influence endothelial survival, inflammatory signalling, and extracellular matrix dynamics.

Taken together, current evidence supports a mechanistic framework in which MSC-EVs suppress neuroinflammation and restore BBB functionality by targeting conserved signalling nodes that couple glial activation to vascular dysfunction. By interrupting reinforcing loops between inflammatory signalling, immune-cell trafficking, and barrier breakdown, MSC-EVs function as active molecular modulators rather than passive trophic agents. Nevertheless, much of the supporting evidence derives from rodent and in vitro systems, and critical gaps remain regarding cell-type-specific EV uptake, dose-response relationships, and temporal requirements for intervention. Addressing these gaps through standardised EV characterisation, longitudinal biomarker profiling of neuroinflammatory and BBB endpoints, and validation in models with higher translational fidelity will be essential to refine this mechanistic framework and guide the rational clinical development of MSC-EV-based therapies.

Neurodegenerative diseases share a pathological module characterised by the tight interdependence of oxidative stress and mitochondrial dysfunction. Under physiological conditions, mitochondria maintain redox homeostasis through tightly regulated reactive oxygen species (ROS) production coupled to adenosine triphosphate (ATP) synthesis and antioxidant buffering. Disruption of this balance leads to excessive ROS accumulation, which damages mitochondrial DNA, lipids, and respiratory chain complexes, further impairing oxidative phosphorylation^[40]. Compromised mitochondrial respiration, in turn, amplifies ROS generation, establishing a self‑reinforcing loop of redox imbalance and bioenergetic failure. This coupling is particularly detrimental in neurons, where high energetic demand, limited antioxidant reserves, and extended cellular architecture restrict compensatory responses. The manifestation of this shared redox-mitochondrial module is shaped by disease- and cell type-specific pathological drivers. In PD, dopaminergic neurons are particularly vulnerable as they rely heavily on mitochondrial oxidative phosphorylation and generate high basal levels of reactive oxygen species because of dopamine metabolism, making them sensitive to mitochondrial impairments^[41]. In AD, mitochondrial dysfunction is tightly coupled to proteotoxic stress, as Aβ and tau directly disrupt mitochondrial dynamics, trafficking, and respiratory function, particularly at synapses^[42]. In ALS, impaired ROS detoxification and mitochondrial damage are prominent in motor neurons expressing mutant SOD1 and TDP-43^[43,44], while HD involves transcriptional dysregulation of mitochondrial biogenesis and energy metabolism^[45,46]. In MS, chronic demyelination disrupts oligodendrocyte-axon metabolic coupling, depriving axons of essential metabolic support and rendering them vulnerable to energy failure, mitochondrial dysfunction, and oxidative injury^[47].

Across these contexts, MSC-EVs have emerged as regulators of this module by intervening at multiple, mechanistically linked control points that collectively decouple oxidative stress from mitochondrial collapse. Rather than acting as simple antioxidants, MSC-EVs deliver defined miRNA, protein, enzyme, and lipid cargo that suppress ROS amplification, restore antioxidant buffering, stabilise mitochondrial integrity and turnover, and thereby prevent mitochondrial stress from progressing toward irreversible neuronal loss [Figure 2]^{[21,32,48-50]}.

A first regulatory axis involves suppression of ROS amplification and reinforcement of antioxidant defence through redox-sensitive transcriptional and signalling programs. In PD cellular and mouse models, UC-MSC-EVs delivering miR-181a-2-3p attenuated ROS accumulation and prevented dopaminergic neuron apoptosis by modulating early growth response-1 (EGR1) via inhibition of the nicotinamide adenine dinucleotide phosphate (NADPH) oxidase 4 (NOX4)-p38 MAPK pathway^[51]. Similarly, trophoblast-derived MSC-EVs enriched in miR-100-5p directly downregulated NOX4, limiting ROS production and engaging the NOX4-nuclear factor erythroid 2-related factor 2 (NRF2) cascade to preserve nigrostriatal integrity and motor performance in MPTP mice^[52]. In a 6-hydroxydopamine (6-OHDA) PD model, human umbilical cord (hUC)-MSC-EVs containing miR-7-5p, miR-125-5p, miR-126-3p, miR-122-5p, and miR-199-3p activated NRF2 signalling, reduced oxidative stress, restored dopamine levels, and improved behavioural outcomes^[53]. Among these, miR-7 directly regulates α-synuclein expression, linking suppression of proteotoxic burden to reduced oxidative load^[54]. Growth-factor cargo can further amplify this response, as brain-derived neurotrophic factor (BDNF)-loaded hUC-MSC-EVs suppressed 6-OHDA-induced apoptosis, activated NRF2, and improved dopaminergic neuron survival and cytoskeletal stability in vitro^[55]. Similar redox-protective reprogramming has been reported in ALS, where BM-MSC-EV-associated miR-467f/466q attenuated p38-MAPK signalling and inflammatory mediator expression in SOD1^G93A microglia and astrocytes^[21,29]. Collectively, these studies position MSC-EV cargo upstream of ROS accumulation, acting on amplification nodes [Figure 2].

Because damaged mitochondria are themselves a major source of secondary ROS, effective antioxidant control is inseparable from preservation of mitochondrial integrity and turnover. This reflects the rate-limiting role of mitochondrial stress in dopaminergic neuron survival. In APP/PS1 AD mice, MSC-EV treatment restored mitochondrial ultrastructure and normalised fission-fusion markers, including COX IV, Tom20, and FIS1, alongside rebalancing of NRF2 pathway components, underscoring reciprocal regulation between redox balance and organelle maintenance^[56]. In PD rat models, UC-MSC-EVs promoted autophagy-dependent neuroprotection, rescuing dopaminergic neurons and normalising dopamine metabolism [3,4-Dihydroxyphenylacetic acid (DOPAC), homovanillic acid (HVA)] and motor behaviour^[57]. In AD, engineered MSC-EVs containing tyrosine phosphatase-2 (SHP2) enhanced mitophagy, supporting selective clearance of damaged mitochondria as an adaptive survival response^[32] [Box 1]. Within this axis, mitochondrial deacetylase sirtuin 3 (SIRT3) emerges as a critical integrator of antioxidant defence and mitochondrial quality control. MSC-EV-mediated SIRT3 activation could potentially enhance SOD2-dependent mitochondrial antioxidant defence, thereby promoting mitochondrial resilience and protecting dopaminergic neurons from oxidative injury ^[58,59] [Figure 2].

In addition to transcriptional and metabolic regulation, MSC-EVs provide enzymatic antioxidant activity that operates on shorter timescales, buffering oxidative stress while longer-term adaptive programs are engaged. Wharton’s Jelly MSC-EVs were able to transfer active catalase to primary rat hippocampal neurons exposed to Aβ oligomers, reducing ROS and preventing synaptic loss^[60]. MSC-EV treatment has been shown to enhance antioxidant capacity in recipient brain tissue, including increased MnSOD (SOD2) and glutathione peroxidases, enzymes capable of detoxifying superoxide and peroxides and thereby stabilising mitochondrial membranes and respiratory complexes^[61]. These enzymatic defences act in parallel with transcriptionally regulated antioxidant systems, notably the NRF2-Keap1 pathway, which coordinates sustained glutathione synthesis and redox homeostasis [Figure 2]. In a rotenone-induced PD rat model, BM-MSC-EVs treatment increased Parkin and DJ-1 levels while reducing α-synuclein accumulation, changes consistent with improved mitochondrial and redox homeostasis^[50]. Reports of MSC-EV-associated mitochondrial components, including mitochondrial transcription factor A (TFAM) and mitochondrial DNA (mtDNA) fragments, further raise the possibility that MSC-EVs contribute to bioenergetic support beyond redox regulation^[62]. However, the extent to which such components functionally integrate into host mitochondrial networks, and the dose and context dependence of these effects, remain incompletely resolved and require systematic validation.

Collectively, current evidence supports a mechanistic framework in which MSC-EVs may mitigate oxidative stress and mitochondrial dysfunction through coordinated regulation of ROS amplification, antioxidant buffering, mitochondrial integrity and turnover, and enzymatic redox control. The predominance of PD models in this literature reflects both the central role of mitochondrial dysfunction in dopaminergic neuron survival and the experimental accessibility of well-established toxin-based models, rather than a selective responsiveness of PD to EV-based interventions. Importantly, EV-mediated modulation of mitochondrial stress response has also been reported across AD, ALS, HD, and MS. This recurring engagement of redox control, mitochondrial quality maintenance, and bioenergetic support suggests that MSC-EVs act on shared cellular stress pathways [Figure 2]. Refining this framework will require causal mapping of EV cargo-target interactions and functional mitochondrial outcomes across disease-relevant cell types, an essential step toward defining therapeutic windows and guiding the clinical translation of MSC-EV-based strategies.

Loss of proteostasis and neurotrophic signalling represent converging pathological mechanisms across neurodegenerative diseases, jointly undermining neuronal survival and repair. Disrupted protein quality control leads to the accumulation of toxic aggregates - including Aβ and tau in AD, α-synuclein in PD, mutant huntingtin (mHTT) in HD, and misfolded SOD1 and TDP-43 in ALS. These aggregates overwhelm proteasomal and autophagic systems, propagate cellular stress, and perturb trafficking and synaptic function. Concurrently, declining neurotrophic input, especially BDNF, glial cell line-derived neurotrophic factor (GDNF), and insulin-like growth factor-1 (IGF-1), reduces synaptic maintenance and regenerative capacity, lowering the threshold for proteotoxic injury^[63,64]. These processes are mutually reinforcing, as proteostatic collapse impairs growth-factor responsiveness while trophic withdrawal weakens autophagy and stress-response programs. Therapeutic strategies that restore protein homeostasis while re-engaging trophic signalling therefore have strong potential for disease modification.

MSC-EVs appear well positioned to target this coupled pathological module. Current evidence suggests that MSC-EVs influence proteostasis through three interrelated mechanisms: enhancing degradative clearance of aggregation-prone proteins, limiting upstream drivers of pathogenic protein production, and stabilising the intracellular protein-folding environment [Figure 3]^[65,66]. These actions simultaneously support neurotrophic signalling and synaptic integrity.

A substantial body of evidence indicates that MSC-EVs facilitate clearance of toxic protein aggregates by delivering proteolytic enzymes and activating intracellular degradation pathways. In AD rodent models, EVs derived from AT- and UC- MSCs delivered active neprilysin (NEP) and insulin-degrading enzyme (IDE), reducing Aβ and improving cognition in APP/PS1 mice^[17,67]. MSC-EVs can also reinforce aggregate clearance by activating signalling programmes that upregulate endogenous proteolytic capacity. BM-MSC-EVs enriched in growth differentiation factor-15 (GDF-15) increased NEP and IDE expression in Aβ42-induced SH-SY5Y cells through activation of the protein kinase B (AKT)/glycogen synthase kinase 3 beta (GSK-3β)/β-catenin pathway, a signalling axis known to regulate transcriptional programmes involved in protein turnover and neuronal survival^[68]. In parallel, sphingosine kinase (SphK)/sphingosine-1-phosphate (S1P) signalling has been shown to enhance degradation of Aβ42 aggregates and preserve synaptic structure and function in AD mice^[69]. Notably, most mechanistic studies of MSC-EV-mediated proteostasis remain focused on Aβ pathology. This represents an important limitation, as tau hyperphosphorylation, aggregation, and intercellular propagation correlate more closely with clinical progression in AD than amyloid burden. Whether MSC-EVs can meaningfully influence tau pathology therefore remains insufficiently resolved.

In parallel with protease-mediated clearance, MSC-EVs enhance intracellular autophagy pathways that enable removal of aggregation-prone proteins. In PD cellular and rodent models, UC-MSC-EV treatment induced autophagic flux by increasing microtubule-associated protein 1 light chain 3-II (LC3-II) levels, facilitated α-synuclein clearance, reduced dopaminergic neuron loss and improved motor outcomes^[57]. 6-OHDA-stimulated SH-SY5Y neuroblastoma cells show that UC-MSC-EVs activate autophagy via AMP-activated protein kinase (AMPK)-mammalian target of rapamycin (mTOR) signalling, thereby reducing apoptotic stress. These effects are recapitulated in vivo, where MSC-EV administration preserves dopaminergic neurons, restores dopamine and its metabolites (DOPAC and HVA), and improves motor performance in PD rat models^[33]. Engineering approaches can amplify these effects, as a hybrid nanocarrier combining MSC-EVs with curcumin (PR-EXO/PP@Cur) reduced α-synuclein oligomers and preserved motor function more effectively than naïve EVs^[33] [Box 1]. At present, it remains unclear whether this benefit reflects enhanced delivery of curcumin alone or synergy with MSC-EV-mediated proteostatic signalling.

Evidence from ALS and HD models suggests that MSC-EVs may also stabilise broader neuronal stress-response programs. In SOD1^G93A neuronal models of ALS, AT-MSC-EVs reduced mutant SOD1 aggregation and restored cAMP response element-binding protein (CREB) phosphorylation and peroxisome proliferator-activated receptor gamma coactivator 1-alpha (PGC-1α) signalling, linking improved proteostasis to mitochondrial support and pro-survival transcriptional programmes^[70]. Consistent with this, our proteomic study in iPSC-derived motor neurons carrying C9ORF72, TDP-43, or FUS mutations identified both shared and mutation-specific ALS signatures and showed that BM-MSC-EVs could reverse proteomic abnormalities across genetic backgrounds, with a functional effect detected in FUS motor neurons in vitro^[71]. Beyond aggregate handling, MSC-EVs also exert regenerative effects on neuronal structure. In SOD1^G93A primary motor neurons, MSC-EV treatment promoted axonal growth, accompanied by enrichment of miRNAs (miR-221-3p, miR-22-3p, miR-125b-3p, miR-145-5p) linked to neuronal differentiation and cytoskeletal regulation^[72,73]. Additionally, an independent study demonstrated that MSC-EV-delivered miR-22 can suppress neuronal pyroptosis by targeting gasdermin-D, thereby preventing inflammatory cell death and indirectly preserving intracellular proteostasis under chronic stress conditions in AD mice^[74]. Together, these findings suggest that MSC-EVs may shift neurons toward a more resilient, repair-competent state [Figure 3].

Beyond promoting degradative clearance, MSC-EVs may also act upstream by suppressing pathogenic protein production, thereby complementing downstream proteolytic and autophagic mechanisms. EV-delivered miR-29c-3p represses beta-site amyloid precursor protein cleaving enzyme 1 (BACE1) via Wnt/β-catenin signalling, thereby reducing amyloidogenic processing^[75]. Additional gene-regulatory strategies have also been reported. For example, engineered MSC-EVs delivering a small interfering RNA targeting fat mass and obesity-associated protein (FTO), an N⁶-methyladenosine (m6A) RNA demethylase implicated in neuronal stress responses, were reported to reduce dopaminergic cell death, potentially via miR-627-5p^[76]. Although promising, such approaches require careful validation of cargo specificity and downstream regulatory networks.

A further mode of MSC-EV action involves stabilising the protein-folding environment through delivery of molecular chaperones and reinforcement of cellular stress-response pathways. Wharton jelly-derived MSC-EVs contain heat shock proteins such as Hsp70^[77]. Hsp70 has been shown to reduce HTT aggregation in cellular HD models, raising the possibility that EV-associated chaperones may influence aggregate burden ^[78,79]. These proteostatic mechanisms likely interact with inflammatory and redox pathways, as suppression of oxidative injury and inflammatory cell death can preserve proteostasis under chronic stress.

MSC-EVs may further contribute to neuronal resilience by reinforcing neurotrophic signalling pathways that support survival, synaptic maintenance, and plasticity. This effect is mediated both through delivery of miRNAs and diverse growth factors, including BDNF, GDNF, IGF-1, and VEGF^[39]. For example, Wharton’s jelly-derived MSC-EVs enriched in miR-29a downregulated histone deacetylase 4 (HDAC4), an epigenetic regulator aberrantly elevated in AD mouse brains, restoring expression of synaptic genes such as Homer1 and Syn2 and improving synaptic plasticity^[77]. Consistent with this, UC-MSC-EV treatment in 6-OHDA models increased microtubule-associated protein 2 (MAP2) expression and reduced hyperphosphorylated tau, indicating improved cytoskeletal stability and neuronal integrity^[57]. Beyond these effects, MSC-EV cargo can directly engage pathways controlling neuronal differentiation and structural plasticity. MSC-EV-mediated delivery of miR-133b, which is markedly downregulated in PD, has been shown to promote neurite outgrowth and neural plasticity through suppression of RhoA and activation of extracellular signal-regulated kinase 1/2 (ERK1/2) and phosphatidylinositol 3-kinase (PI3K)/Akt signalling, while also regulating Pitx3, a transcription factor critical for dopaminergic neuron development and maintenance^[80,81]. Similarly, MSC-EVs carrying miR-124 and miR-145 have been shown to drive maturation of human neural progenitors and increase glutamate transporter expression in neurons and astrocytes, supporting functional network integration and excitatory homeostasis^[82].

At the circuit level, MSC-EVs can translate these molecular effects into restoration of synaptic function. In APP/PS1 mice, MSC-EV treatment restored CA1 pyramidal neuron excitability disrupted by Aβ-associated calcium transient abnormalities. In AD mice, EVs from hypoxia-preconditioned MSCs reduced Aβ oligomers while increasing growth-associated protein-43 and synapsin-1 and improving memory performance, consistent with coupled effects on aggregate stress and synaptic repair^[31] [Box 1]. In oligodendrocyte progenitor cells (OPCs) derived from EAE mice, UC-MSC-EV delivery of miR-23a-3p activated T-box brain protein 1 (Tbr1)/Wnt and PI3K/Akt signalling to promote oligodendrocyte differentiation and myelin protein expression, indirectly preserving neuronal proteostasis by reducing axonal metabolic stress^[66].

Collectively, these findings support a unified mechanistic framework in which MSC-EVs restore proteostasis and neurotrophic support through tightly coupled actions: promoting autophagic and proteolytic clearance, suppressing pathogenic protein production, stabilising folding environments, and re-engaging survival and plasticity signalling. Rather than acting as passive carriers of growth factors, MSC-EVs function as dynamic regulators of neuronal resilience, integrating trophic signalling with protein quality control across disease contexts.

MSC-EVs are increasingly investigated as therapeutics for neurodegenerative disease, yet how molecular cargo differences across MSC sources translate into distinct or overlapping therapeutic mechanisms remains poorly understood. Comparative proteomic and transcriptomic analyses reveal that MSC-EVs derived from BM, AT, and UC share a conserved set of paracrine factors enriched in immunomodulatory, antioxidant, and trophic factors, alongside differences in the relative abundance of specific cargo classes^[85-87]. These differences reflect biological variation between tissue sources rather than discrete functional identities and are best interpreted as quantitative shifts in cargo composition that may influence the relative engagement of mechanistic axes.

Consistent with this view, BM-MSC-EVs show relative enrichment in proteins involved in intracellular trafficking, proteasome-associated machinery, as well as synaptic regulators such as ADAM10, ADAM9 and Notch2^[86]. AT-MSC-EVs tend to show stronger representation of secretory and immunomodulatory pathways, and oxidative stress response programmes^[86,88]. UC-MSC-EVs display higher relative abundance of proteins such as plasminogen activator inhibitor (PAI)-1 and 14-3-3ζ (YWHAZ), associated with Wnt signalling, extracellular matrix (ECM) remodelling and vascular integrity^[88,89]. Importantly, these molecules are not exclusive to any single MSC source but differ in relative abundance, and their functional implications should therefore be considered probabilistic rather than deterministic.

Such cargo differences are often interpreted as aligning with distinct mechanistic emphases, as illustrated by studies in AD. In AD cellular and rodent models, MSC-EVs have most often been linked to synaptic maintenance and neuronal survival, consistent with their ability to restore synaptic plasticity‑related gene expression and improve cognitive deficits^[77]. AT-MSC-EVs have been prominently associated with extracellular proteostasis through delivery of Aβ-degrading enzymes such as neprilysin, reducing amyloid burden in vitro and in vivo settings^[67]. UC-MSC-EVs have frequently been studied in relation to immune and neurovascular interfaces, reprogramming microglia toward a more phagocytic state and promoting Aβ clearance in vivo, consistent with their receptor-, Wnt-, and ECM-related cargo^[86,89]. However, these patterns largely reflect differences in experimental focus and study design rather than controlled source comparisons within identical models. As a result, they should be interpreted as mechanistic tendencies, not evidence of source superiority.

Despite quantitative differences in cargo composition, preclinical studies report broadly beneficial effects from MSC-EVs across sources^[90], consistent with the possibility that the shared paracrine core may be sufficient to confer therapeutic impact in commonly used preclinical animal models^[86,91]. However, the field still lacks rigorous comparative studies controlling for MSC origin, EV manufacturing, dosing, route, and outcome measures, precluding designation of an optimal MSC source for neurodegenerative applications.

Taken together, available evidence indicates that MSC tissue source influences the relative abundance of bioactive proteins and regulatory RNAs packaged into EVs, thereby modulating the balance of pathways engaged in recipient cells. These differences do not constitute discrete therapeutic identities, nor do they justify designation of a superior source for neurodegenerative applications. Rather, they suggest that MSC origin modulates the balance of shared regulatory mechanisms within a conserved paracrine framework. The central implication is therefore not source selection as a binary choice, but recognition that tissue origin may fine-tune module engagement. Resolving whether such tuning is clinically meaningful will require systematic head-to-head comparisons conducted under harmonised manufacturing and delivery conditions.

The therapeutic efficacy of MSC-EVs in neurodegeneration depends not only on their cargo composition but also on whether EVs and/or their bioactive cargo achieve sufficient exposure at disease-relevant sites. Neurodegenerative disorders are spatially heterogeneous and characterised by dynamic neurovascular dysfunction, glial activation, and altered immune trafficking. In this context, administration route and pharmacokinetic behaviour primarily shape systemic clearance, peripheral sequestration, and the likelihood of CNS exposure. Although MSC-EV-associated signals have been detected beyond the BBB, their degree of CNS exposure is strongly influenced by vascular state, delivery route, and disease context^[90]. Brain delivery constraints and methodological caveats in EV tracking have been reviewed extensively and remain central to interpreting the literature^[91].

Across studies, intravenous (IV) administration of EVs results in rapid clearance by the mononuclear phagocyte system, with dominant uptake in liver and spleen and only a minute fraction accessing the CNS^[92-94]. Despite this limited brain accumulation, IV-delivered MSC-EVs consistently produce molecular and functional benefits in rodent models of AD, PD, ALS, and MS^{[20,56,57,95]}. This mismatch between limited brain accumulation and measurable benefit challenges the assumption that widespread parenchymal penetration is a prerequisite for therapeutic efficacy. One explanation is that neurodegenerative pathology is associated with regional BBB dysfunction and neurovascular inflammation, which may increase permissiveness to circulating MSC-EVs compared with healthy brain. A second possibility is that many MSC-EV effects may occur at neurovascular and immune interfaces, where modulation of endothelial cells and perivascular macrophages can propagate secondary effects into neural networks without requiring extensive neuronal uptake^[96,97]. Whether similar exposure-effect relationships hold in humans remains uncertain and defining dose-exposure-response relationships represents a key translational gap. Systemic MSC-EV administration, delivered as either single or repeated doses, has improved behavioural and disease-associated inflammatory and pathological outcomes in AD, PD, and ALS mice^[20,53,56], but whether this reflects greater CNS exposure, prolonged peripheral immunomodulation, or both is not yet resolved.

Intranasal (IN) administration has been explored as a non-invasive strategy to bypass hepatic clearance by exploiting olfactory and trigeminal pathways. In rodents, IN-delivered UC-MSC-EVs labelled via amine-reactive conjugation of EV surface proteins were detected in the hippocampus, cortex, and olfactory bulbs, and are taken up by neurons, astrocytes, and microglia^[98]. Perets et al. used computed tomography combined with gold nanoparticle labelling and reported selective accumulation of IN-administered BM-MSC-EVs within inflamed or degenerating regions in AD and PD models, with detectable signal persisting for up to 96 h^[99]. Moreover, functional studies reporting reductions in Aβ burden and neuroinflammation after IN administration often do not provide robust quantification of intact vesicle accumulation in brain parenchyma, despite observing behavioural improvement^[18,100]. In EAE models, for example, signal has been reported predominantly in peripheral lymphoid tissues rather than CNS tissue^[101]. Together, these data support functional benefit in rodent systems while leaving unresolved the extent to which therapeutic effects depend on direct parenchymal EV uptake versus indirect immune or neurovascular mechanisms.

However, translational scaling introduces substantial anatomical and physiological constraints in large animals. In pig-tailed macaques, IV-administered MSC-EVs labelled with a membrane-anchored bioluminescence reporter (palmGRET) exhibited longer circulation times and low but measurable cerebrospinal fluid (CSF) levels yet remained predominantly sequestered in liver and spleen^[102]. The same study further demonstrated negligible nose-to-brain delivery following IN administration, with a substantial proportion of administered EVs recoverable from the nasal cavity^[102]. These findings underscore mucociliary clearance and epithelial barriers as dominant constraints on CNS access in primates. Positron emission tomography (PET)/magnetic resonance imaging (MRI) studies in rhesus macaques using radio-labelled EVs showed that intrathecal (IT) administration achieved the highest CNS exposure, IV delivery produced low but detectable brain signal, and IN delivery resulted in negligible CNS distribution^[103]. These findings indicate that rodent “nose-to-brain” efficiency does not scale directly to primates, likely due to differences in nasal anatomy, mucociliary clearance, and epithelial barriers. Whether therapeutically sufficient CNS exposure can be achieved through IN delivery in humans remains unresolved.

Interpretation of EV biodistribution data is further complicated by methodological limitations. Many rodent studies rely on lipophilic dyes such as PKH to visualise EV brain uptake. While useful for in vitro and short-term tracking, these dyes can transfer to host membranes or persist following EV degradation, confounding attribution of signal to intact EVs^[39,104]. Radiolabelling and genetic reporters improve specificity but do not fully resolve whether detected signal represents intact vesicles, phagocytosed material, or redistributed cargo^[92]. Moreover, most studies do not distinguish between the fate of the EV membrane and that of its functional cargo after cellular uptake. This distinction is particularly relevant in the CNS, where limited EV entry may still influence network-level inflammatory and metabolic signalling. Dose-exposure-response relationships remain poorly defined, with many studies employing supraphysiological EV doses without establishing minimal effective thresholds or persistence within CNS compartments. Immune responses add an additional layer of complexity, as repeated IV administration of xenogeneic EVs in macaques accelerates clearance^[102], an effect rarely quantified in rodent neurodegenerative models.

Taken together, current evidence supports three pragmatic conclusions. First, native MSC-EVs show low CNS exposure after systemic delivery in primates, and the biological significance of this exposure remains to be established in neurodegenerative contexts. Second, administration route strongly determines systemic clearance and relative CNS access, with IT delivery providing the most reliable CNS exposure but limited by invasiveness, while IN delivery shows limited scalability to primates. Third, engineering strategies provide proof-of-principle that targeting can be shifted^[9]. Rabies virus glycoprotein (RVG)-modified MSC-EVs improve neuronal uptake^[31], and ligand-guided approaches redirect EV tropism toward injured brain regions in rodent models^[37]. Additionally, effective IN delivery in humans will likely require strategies that overcome mucociliary clearance and epithelial barriers. Approaches such as transient mucolytic pretreatment or surface modification remain insufficiently evaluated for MSC-EVs in vivo^[105,106]. From a translational perspective, the central objective is therefore not simply to maximise brain uptake, but to define which level of exposure is sufficient for module regulation and to establish quantifiable robust biomarkers of CNS target engagement capable of anchoring dose selection and delivery strategy.