Mesenchymal Stem Cells and Extracellular Vesicles for Neurodegenerative Diseases: Therapeutic Advances and Challenges Authors: Luming Pan, Baocai Liu, Jingjia Xu, Xupeng Mu, Lianlian Fan Published: 2026 ABSTRACT AbstractMesenchymal stem cells (MSCs), as a type of adult stem cells, exhibit robust self-renewal, multi-lineage differentiation, paracrine and immunomodulatory capacities, demonstrating broad application prospects in the treatment of neurodegenerative diseases. This review systematically summarizes the mechanisms of action, therapeutic advances and comparative analyses of various MSCs and their derived extracellular vesicles (EVs) in Alzheimer’s disease (AD), Parkinson’s disease (PD) and amyotrophic lateral sclerosis (ALS), and discusses the optimization effects of gene modification and pretreatment on therapeutic efficacy. This article not only highlights the advantages of various MSCs and their corresponding EVs, but also provides unique insights into their differentiated therapeutic potential and mechanism of action, which have not been fully elucidated in previous studies. Meanwhile, although current research results are encouraging, this paper also critically points out that existing studies suffer from insufficient elucidation of mechanisms and lack of large-scale clinical trials. Finally, the article prospects future directions of MSC-based therapeutic strategies, including mechanism deepening, treatment optimization and standardization system construction, to promote their translation into clinical application. However, most of the current evidence is still preclinical, and the recognized clinical efficacy in humans is still limited. This is a narrative review. Literature was screened from Web of Science and PubMed by thematic relevance and research quality, without systematic review protocol or meta-analysis.Keywords: mesenchymal stem cells, neurodegenerative diseases, extracellular vesicles, genetic engineering modification, clinical application Introduction Neurodegenerative diseases are a class of neurological disorders characterized by progressive loss of neuronal structure or function, mainly involving the brain, spinal cord and other central nervous systems. They mainly include AD, PD, ALS, which represent the most prevalent, disabling, and fatal subtypes of neurodegenerative disorders with distinct clinical and pathological features. It is estimated that with the progress of global aging, over 276 million people are affected by neurodegenerative diseases, which are the second leading cause of death worldwide.1 Their main pathologies include protein misfolding, neuroinflammation and neuronal loss, leading to a series of clinical symptoms such as cognitive dysfunction and movement disorders.2 Current therapies can only delay disease progression and cannot reverse the course.3 Therefore, there is an urgent need to develop therapies that can intervene in the essence of the disease. Compared with neural stem cells (NSCs), embryonic stem cells (ESCs), and induced pluripotent stem cells (iPSCs), which are subject to ethical controversies, limited sources, high risk of immune rejection, and limitations in tumorigenicity, MSCs have become the preferred seed cells for regenerative therapy of neurodegenerative diseases due to their unique advantages.4 MSCs are easy to obtain, have low immunogenicity, strong differentiation ability and can secrete a large number of cytokines, becoming the first choice of cells for disease-modifying therapy in regenerative medicine. In addition, compared with ESCs and iPSCs, MSCs have a lower risk of tumor formation, and their isolation and in vitro amplification protocols are relatively simple and mature, laying a solid foundation for large-scale clinical applications. However, MSCs derived from different tissues such as bone marrow, tissue or umbilical cord differ in proliferation ability, differentiation tendency, secretion profile and immunomodulatory intensity. These differences directly affect their therapeutic effect and application potential for neurodegenerative diseases. Therefore, different diseases require different cell types. Although a large number of studies have explored the therapeutic effects of MSCs in neurodegenerative diseases, most studies still remain at the level of a single cell type, lacking systematic horizontal comparative studies. At the same time, research on enhancement measures such as EVs, gene modification and pretreatment also lacks systematic review. More critically, there are still gaps between basic research and clinical application such as unclear mechanisms, limited models and insufficient clinical evidence, and existing reviews rarely face these pain points directly. Previous work has rarely systematically elucidated the source specificity and limitations of each MSC subtype when applied to different neurodegenerative diseases. To address these critical research gaps and overcome the deficiencies of existing reviews, this narrative review will provide a systematic, comparative and critical analysis. It first compares the biological characteristics of MSCs from different sources, then deeply elaborates and compares their therapeutic mechanisms and research progress in AD, PD and ALS. Subsequently, it evaluates optimization strategies such as EVs therapy, gene modification and pretreatment. This paper focuses on critically pointing out the core limitations and challenges of current research. Finally, it proposes clear future research directions and lays the foundation for conducting standardized clinical trials. Classification and Characteristics of MSCs Bone Marrow-Derived MSCs (BM-MSCs) BM-MSCs are non-hematopoietic multipotent stem cells isolated from bone marrow stroma. BM-MSCs can differentiate across lineages into osteoblasts, chondrocytes, adipocytes, etc.5 In terms of immunomodulation, they can secrete cytokines such as prostaglandin E2 (PGE2) to inhibit excessive activation of immune cells. Li et al found that in a liver transplantation animal model, HO-1-modified BM-MSCs achieved immunosuppression and reduced allograft rejection by inhibiting NK cell activity and expanding regulatory T cell (Treg) populations, while decreasing interleukin (IL) and tumor necrosis factor (TNF) levels.6 Moreover, BM-MSCs can secrete large amounts of bioactive molecules such as vascular endothelial growth factor (VEGF), fibroblast growth factor (FGF) and exosomes, promoting angiogenesis and tissue repair.7 With these characteristics, BM-MSCs can be applied in bone injury repair, neurodegenerative diseases and cardiac repair. However, cell acquisition requires bone marrow aspiration, and cell proliferation and differentiation capacity significantly decline with donor age,8 while post-transplantation survival rate is also limited, restricting their clinical application.9 Adipose-Derived MSCs (AD-MSCs) AD-MSCs are harvested from discarded tissue after liposuction or lipectomy without requiring bone marrow puncture. The cell yield per gram of adipose tissue far exceeds that of bone marrow, and their in vitro expansion capacity is pronounced.10 Like BM-MSCs, AD-MSCs retain multi-lineage differentiation potential and can be induced in vitro to differentiate into osteocytes, chondrocytes, and adipocytes; nevertheless, their differentiation propensities diverge markedly. Zhong et al noted that, because the activation states of peroxisome proliferator-activated receptor (PPAR) and WNT signaling pathways governing cartilage and bone development differ between AD-MSCs and BM-MSCs, AD-MSCs display stronger osteogenic and adipogenic differentiation yet weaker chondrogenic differentiation than BM-MSCs.11 Najar et al further reported that human AD-MSCs suppress lymphocyte proliferation in a dose-dependent fashion, an effect surpassing that of BM-MSCs under identical conditions.12 Quaedackers et al documented that AD-MSCs directly engage activated T-cell, B-cell, and NK-cell subsets, modulating their functional status and thereby exerting immunoregulatory actions. Simultaneously, AD-MSCs secrete VEGF, bFGF, and brain-derived neurotrophic factor (BDNF), furnishing a supportive microenvironment for tissue repair and regeneration.13 Human Umbilical Cord Mesenchymal Stem Cells (hUC-MSCs) hUC-MSCs are obtained from Wharton’s jelly of the umbilical cord, show long-term expansion ability in vitro and low immunogenicity. Under specific induction conditions they can differentiate into various cells of all germ layers, exhibit a wider multi-directional differentiation capacity than MSCs from other sources, and offer a cellular basis for multiple tissue reconstruction. Research indicates that they have active paracrine functions, continuously secrete growth factors, chemokines and exosomes, affect inter-cell communication and modulate the cellular microenvironment.14 In addition, hUC-MSCs can promote the differentiation of Th17 cells into Treg cells via the transforming growth factor-β (TGF-β) pathway, thus suppressing inflammation.15 This provides a new mechanism for treating immune-related diseases. With their low immune rejection, multi-directional differentiation potential, bidirectional immunomodulation and paracrine properties, hUC-MSCs have been used in the therapeutic exploration of inflammatory diseases such as autoimmune hepatitis, cardiovascular system diseases and neurodegenerative diseases.15–18 In-vivo Therapeutic Strategies of Various MSCs Treatment of Alzheimer’s Disease (AD) Therapy of BM-MSCs Against AD Studies have shown that transplantation of BM-MSCs possesses potential therapeutic value for AD; their protective effect first relies on cytokines released within a short time after entering the brain, which can induce host degrading enzymes to clear amyloid-β (Aβ), suppress neuro-inflammation, antagonize oxidative stress and promote neural repair together with synaptic remodelling. Abnormal deposition of Aβ is one of the core pathological features of AD. Gu et al found that BM-MSCs up-regulate total amyloid precursor protein (APP) expression in the brain of APP/PS1 mice and guide its metabolism through the non-amyloidolytic pathway, thereby reducing the generation of pathogenic Aβ.19 However, whether this process is accomplished by activating α-secretase or inhibiting β-secretase remains inconsistent, and whether it can be repeated in sporadic or other transgenic models is still lacking conclusion. The same study confirmed that BM-MSCs can effectively decrease cerebral Toll-like receptor 2 expression, inhibit nuclear factor-κB (NF-κB) nuclear translocation and reduce the induction of inducible nitric oxide synthase (iNOS), thereby suppressing excessive NO production; this anti-inflammatory effect, on the one hand, alleviates intracerebral inflammation and, on the other hand, indirectly attenuates peroxynitrite-mediated lipid peroxidation. Subsequently, decreased NF-κB activity relieves the inhibition of nuclear factor-erythroid 2-related factor 2 (Nrf2), activates the Nrf2/HO-1 signalling pathway, strengthens cellular antioxidant defence and partially restores redox balance. Meanwhile, Yan et al reported that tail-vein BM-MSCs can effectively reduce peripheral blood levels of IL-1 and other pro-inflammatory cytokines in AD model mice, indicating that BM-MSCs may form a synergistic anti-inflammatory-antioxidant cascade by modulating both central-local and peripheral systemic inflammation, thus furnishing a microenvironment for neuronal survival and synaptic remodeling. But this conclusion needs to be confirmed through combined administration.20 Regarding neural regeneration and synaptic remodelling, BM-MSCs continuously secrete insulin-like growth factor 1 (IGF-1), VEGF, etc., inhibit Caspase-3 activity, up-regulate expression of apoptosis-inhibiting protein family members and decrease neuronal apoptosis. Their secreted cytokines can also elevate Akt phosphorylation levels and restore neuronal function through the Akt/IAPs pathway, thereby achieving therapeutic effects.21 Another study showed that BM-MSCs up-regulate the expression of neural progenitor markers such as nestin and SOX-2 and trigger endogenous neural repair and regeneration.22 Finally, Bae et al pointed out that BM-MSCs significantly increase the expression of synaptic transmission-related proteins dynamin 1 and synapsin 1 and improve synaptic transmission.23 After co-culture of BM-MSCs with Aβ-injured neural cells, BM-MSCs markedly ameliorate the micro-environment of Aβ-injured neural cells by secreting anti-inflammatory factors such as IL-6, IL-10 and TGF-β; simultaneously they influence cell survival, metabolism and regeneration through regulating the mechanistic target of rapamycin (mTOR) pathway, AMP-activated protein kinase (AMPK) pathway and Wnt3/β-catenin signalling, and significantly elevate telomerase activity to exert neuro-protective effects.24 These findings are highly consistent with the aforementioned functions of BM-MSCs in modulating APP metabolism, inhibiting neuro-inflammation, antagonizing oxidative stress and promoting neural regeneration, jointly revealing the possible mechanistic basis of BM-MSCs as a cellular therapeutic strategy for AD. In recent years, clinical research on BM-MSC therapy for AD remains at a preliminary stage. A Phase I clinical trial in mild AD patients showed that intravenous infusion of allogeneic BM-MSCs possesses good safety and tolerability.25 In terms of efficacy, the therapy displayed a trend of delaying disease progression; cognitive scale scores of some patients improved, and daily-activity capacity was significantly better than that of the placebo group. Imaging analyses revealed that treatment slowed whole-brain atrophy and manifested protective effects on hippocampal structure. In addition, serum levels of soluble TIE2 (sTIE2) decreased in the treatment group, suggesting that the effect may be related to improved vascular endothelial function. However, the study was limited by small sample size and short observation period, and the efficacy of BM-MSCs in moderate-to-severe AD patients remains to be verified in follow-up studies. Therapy of AD-MSCs Against AD The intervention effect of AD-MSCs on AD is mainly concentrated in three aspects, mainly including promoting the clearance of Aβ, regulating neuroinflammation and exerting neuroprotection. After intracerebral injection of AD-MSCs into AD model rats, Nabil M’s team observed elevated neprilysin expression and up-regulated Sirtuin1 (SIRT1) levels. SIRT1 was negatively correlated with Aβ levels, indicating that SIRT1 may become a potential target for AD therapy.26 Ma et al further confirmed that AD-MSC transplantation reduced Aβ deposition in the hippocampus and cortex of AD model mice, up-regulated the expression of various Aβ-degrading enzymes including insulin-degrading enzyme, and enhanced Aβ clearance capacity.27 Regarding neuro-inflammation modulation, AD-MSCs can not only effectively suppress the expression of pro-inflammatory cytokines such as IL-1β, but also promote the polarization of microglia toward a protective “alternatively activated” phenotype, manifested by decreased pro-inflammatory cytokine expression and increased anti-inflammatory cytokine and alternative activation marker expression, thereby alleviating neuro-inflammation.27 In addition, AD-MSCs possess the potential for neuro-protection and regeneration promotion. They decrease neuronal apoptosis by down-regulating the pro-apoptotic protein Bax and up-regulating the anti-apoptotic protein Bcl-2, while promoting the expression of neurotrophic factors and improving synaptic plasticity.26 It is noteworthy that the therapeutic effect of AD-MSCs may be dose-dependent. In the study by Babaei H et al, low-dose AD-MSCs improved cognitive deficits by enhancing anti-oxidant, anti-inflammatory and anti-apoptotic effects, whereas the effect of the high-dose group was weakened, possibly related to local nutrient competition after cell transplantation or cellular senescence caused by long-term culture.28 This finding provides an important basis for optimizing AD stem cell therapy strategies. Although animal experiments have confirmed that AD-MSCs can reduce cerebral Aβ burden and suppress neuro-inflammation, systematic clinical trials to prove their safety and efficacy are still lacking. Because of their wide source and strong proliferation capacity, they provide a basis for subsequent clinical trials. Therapy of hUC-MSCs Against AD hUC-MSCs mainly exert their therapeutic effects on AD through paracrine mechanisms. Wang utilized gene-silenced strains to demonstrate that hUC-MSCs secrete multiple cytokines that act synergistically: hepatocyte growth factor is responsible for clearing Aβ and microtubule-associated protein tau, while BDNF inhibits excessive microglial activation and improves synaptic function.14 Jia et al further clarified that hepatocyte growth factor (HGF) activates its receptor c-Met, initiates the downstream AKT-GSK3β signalling pathway, thereby suppressing glycogen synthase kinase-3β (GSK3β) activity, down-regulating tau hyper-phosphorylation and ameliorating neurofibrillary tangle pathology.29 In the oxidative-stress segment, Yang et al intravenously transplanted hUC-MSCs into AD model mice and found that malondialdehyde content in the hippocampus decreased while superoxide dismutase (SOD) activity simultaneously increased, lowering cerebral oxidative-stress levels.30 hUC-MSCs can exert anti-inflammatory effects while treating AD. Xie et al observed that hUC-MSC treatment increased IL-10 expression in AD mice and reduced the activation of pro-inflammatory microglia, exerting an anti-inflammatory effect.31 Another study pointed out that hUC-MSC treatment can inhibit abnormal activation of hippocampal astrocytes and restrict the release of pro-inflammatory cytokines such as TNF-α.32 The therapeutic effects of hUC-MSCs even extend to processes such as cell aging. After intravenous transplantation of hUC-MSCs, the expression of senescence-related genes in mouse brain tissue changed significantly: expression of silent information regulator 2 and proliferating cell nuclear antigen was up-regulated, whereas expression of the tumour-suppressor gene p53 and its downstream p21 gene was down-regulated; by inhibiting the p53/p21 pathway, cell-cycle arrest was alleviated, endogenous repair and anti-aging capacity were enhanced, cognitive ability was improved and aging was delayed.33 Considering that hUC-MSCs possess primitive characteristics and multi-directional differentiation potential, some studies pre-induced them into a neuron-like phenotype before transplantation. This approach further reduced Aβ deposition and soluble Aβ levels in mice and increased synapsin-1 expression, a mechanism related to M2-like microglial polarization that plays an important role in synaptic maturation and neurotransmitter release.34 Comparison of Therapeutic Mechanisms of the Three Cells in AD Although the three kinds of mesenchymal stem cells can all intervene in AD at multiple targets, their specific mechanisms are different, and their acquisition difficulty and differentiation potential are also different. The synergy and complementarity of the three mechanisms provide a basis for the development of combined treatment strategies. Regarding Aβ metabolism, BM-MSCs focus on upstream regulation, guiding APP to metabolize through the non-amyloidolytic pathway to reduce Aβ production. AD-MSCs focus on downstream clearance, clearing Aβ by up-regulating the expression of degrading enzymes. hUC-MSCs show the ability of collaborative intervention, which can not only secrete HGF to cooperatively clear Aβ, but also intervene in tau protein hyper-phosphorylation by regulating GSK3β activity (Figure 1). Regarding immune regulation, BM-MSCs can simultaneously down-regulate pro-inflammatory factors in both the brain and the periphery; while AD-MSCs mainly act on the central nervous system, inducing microglia to polarize towards a neuroprotective phenotype; hUC-MSCs increase the level of IL-10, inhibiting excessive activation of pro-inflammatory microglia and astrocytes (Figure 2). MSCs modulate Aβ/tau pathology via distinct nodes. Left: BM-MSCs promote non-amyloidogenic APP cleavage, reducing Aβ generation. Right: AD-MSCs enhance Aβ clearance; hUC-MSCs secrete HGF to clear Aβ and inhibit GSK-3β, reducing tau hyperphosphorylation. Created with BioRender.com. MSCs exert anti-neuroinflammatory effects in AD. In the brain (left), BM-MSCs inhibit TLR2/NF-κB and activate Nrf2/HO-1, reducing iNOS/NO and lipid peroxidation; AD-MSCs promote M2 microglia and inhibit IL-1β; hUC-MSCs suppress M1 microglia and secrete IL-10. The upward arrow indicates an increase in protein expression, while the downward arrow indicates a decrease in protein expression. In peripheral blood (right), BM-MSCs reduce circulating IL-1, IL-2, and TNF-α. Created with BioRender.com. In terms of antioxidant defense, BM-MSCs can activate the classic defense pathway of Nrf2/HO-1; hUC-MSCs enhance the activity of SOD and reduce the content of malondialdehyde; although experiments have demonstrated that AD-MSCs have the ability to alleviate oxidative stress, the precise mechanism has not yet been clarified. At the level of nerve repair and anti-aging, BM-MSCs secrete neurotrophic factors to activate AKT and other pathways, up-regulate neural precursor cell markers, and stimulate endogenous repair. AD-MSCs finely regulate the Bax/Bcl-2 balance to reduce neuronal apoptosis. hUC-MSCs rely on the down-regulation of p53/p21 to delay cell senescence. Except for the differences in the mechanisms mentioned above, their effects on synapses have similar characteristics. BM-MSCs increase the expression level of synaptic transmission-related proteins, AD-MSCs can promote the expression of neurotrophic factors, and hUC-MSCs-induced neuron-like cells can increase the expression of synapsin-1, jointly repairing the damaged neural connections in AD (Figure 3). BM-MSCs elevate Synapsin1 and dynamin1. hUC-MSCs first differentiate into neuron-like cells and then up-regulate Synapsin1; AD-MSCs secrete neurotrophic factors to drive synaptic remodeling-together enhancing synaptic transmission in AD mice. The dashed arrow indicates indirect, paracrine-mediated effects, whereby AD-MSCs secrete neurotrophic factors into the microenvironment, which in turn act on synaptic structures to promote plasticity and survival. Created with BioRender.com. From the perspective of clinical translation, the three mesenchymal stem cell therapies have complementary advantages and disadvantages. BM-MSCs have the most complete data, but bone marrow puncture and age-related decline limit their wide application. AD-MSCs are easy to obtain, proliferate rapidly, and have better cryopreservation tolerance than BM-MSCs, but their efficacy is dose-dependent and lacks clinical trials. hUC-MSCs have primitive characteristics, strong paracrine function, and low immunogenicity, but their clinical translation is still in the early stage, and long-term safety and efficacy need to be verified. In the future, combined treatment strategies can be explored, such as using AD-MSCs in the early stage of the disease to quickly clear Aβ, and then using hUC-MSCs for anti-aging and synaptic remodeling. This strategy needs to be verified by pre-clinical and clinical trials. Parkinson’s Disease Therapy Therapy of Bone-Marrow-Derived MSCs Against PD Within PD-repair strategies BM-MSCs improve dopaminergic function via secreted factors, homing migration and local niche remodelling. In a D-galactose-induced brain-ageing rat model, G. El-Akabawy et al observed that BM-MSCs transplantation up-regulated tyrosine hydroxylase (TH) and vesicular monoamine transporter-2 (VMAT-2) expression, migrated and differentiated into TH-positive dopaminergic neuron-like cells, lowered striatal malondialdehyde levels and restored BDNF and other neurotrophic factors.35 In 6-OHDA PD rats, BM-MSCs likewise restored striatal dopamine and its precursors, elevated SOD activity, down-regulated IL-6 and other inflammatory cytokines and alleviated neuro-inflammation. Increased TH expression indicated recovered dopaminergic function together with enhanced neurotrophic support and symptom amelioration.36 Homing assays showed that tail-vein-injected BM-MSCs migrated to injured brain regions and up-regulated the neural-progenitor marker nestin, implying endogenous repair potential.37 Chen et al compared delivery routes: direct BM-MSC grafting into the substantia nigra of PD rats more effectively promoted local dopaminergic neuron recovery; grafted cells differentiated into neuronal progenitors and mature neurons, suggesting that nigra-targeted transplantation provides a more direct micro-environmental support for dopaminergic regeneration.38 Recent clinical trials have explored allogeneic BM-MSCs for PD. Schiess et al used single, dose-escalating intravenous infusions and verified that allogeneic BM-MSCs are safe and well tolerated in mild-to-moderate PD; the highest dose showed significant therapeutic potential, with decreased peripheral pro-inflammatory cytokines, increased brain-derived neurotrophic factor, improved basal-ganglia perfusion and sustained motor-score improvement.39 Another trial implanted BM-MSCs bilaterally into the sub-ventricular zone; again no serious adverse events were observed and early-stage patients exhibited the clearest motor gains. MR-tractography indicated partial neural-pathway repair, implying potential neuro-protection. However, in patients with advanced PD, the clinical efficacy of unilateral subventricular zone transplantation is relatively limited. Only a small number of patients show improvement in Unified Parkinson’s Disease Rating Scale (UPDRS) scores, with limited overall symptom relief, and it fails to reduce the dosage of dopaminergic drugs.40 Current trials are, however, limited by small sample sizes, lack of randomised double-blind placebo controls and short follow-up, and efficacy in advanced stages remains low. Large-scale phase-II randomised controlled studies are required to confirm the efficacy and long-term safety of BM-MSCs for PD. Therapy of Adipose-Derived MSCs Against PD The therapeutic potential of AD-MSCs for PD has been documented in multiple pre-clinical studies. Yildirim S et al grafted AD-MSCs into the MPTP-induced PD mouse model and effectively increased the number of TH-positive neurons in the substantia nigra, accompanied by up-regulation of BDNF and elevation of IL-10 and TGF-β, indicating combined neurotrophic and anti-inflammatory actions.41 Beyond central neuro-protection, AD-MSCs also modulate the systemic immune system. Clinical observations show that the proportion of Th17 cells is elevated and Treg cells are impaired in the peripheral circulation of early-stage PD patients; this immune imbalance correlates with disease severity and can aggravate central neuro-inflammation.42–47 Bi found that AD-MSCs inhibit Th17 differentiation and the expression of its nuclear transcription factor RORγt, raise the Treg ratio and the secretion of its effector cytokine IL-10, reshape the Th17/Treg balance and ameliorate inflammation.48 Proteomics studies have shown that AD-MSCs in the damaged microenvironment can secrete large amounts of Pertraxin 3 (PTX3), blocking the extrinsic apoptotic pathway and reducing the degeneration of DA neurons, thereby exerting neuroprotective effects. This research provides a theoretical basis for explaining its efficacy and developing cell-free therapies.49 At present, clinical evidence for AD-MSC therapy in PD still comes from small-sample trials. Repeated intravenous infusion of autologous AD-MSCs in elderly patients is safe and feasible, and motor function and quality of life show small-to-moderate improvement, indicating clinical potential. However, due to the small sample size and short design period, the efficacy requires confirmation through large-scale randomized studies.50 Therapy of Human Umbilical Cord-Derived MSCs Against PD Pre-clinical studies usually pre-induce hUC-MSCs into a dopaminergic phenotype before transplantation in order to achieve direct cell replacement. Weiss et al discovered earlier that under specific neural induction conditions, hUC-MSCs can differentiate into catecholaminergic neurons. Although transplantation improved the motor function of PD rats, no long-term surviving cells were detected, suggesting that its therapeutic effect mainly relies on paracrine effects rather than cell replacement.51 Subsequent studies optimized induction scheme on this basis. Liu et al utilized a combination of small molecules to efficiently transform hUC-MSCs into dopaminergic neurons. The resulting cells not only expressed pan-neuronal markers such as neuron-specific class III beta-tubulin (TUBB3), but also specifically and highly expressed key dopaminergic markers such as TH, actively secreted dopamine, and exhibited electrophysiological characteristics of mature neurons. After transplantation into the striatum of PD mice, it increased local dopamine concentration and improved motor dysfunction.52 To compare the therapeutic effects of cells at different stages of differentiation, Zhao et al transplanted hUC-MSCs and induced dopaminergic neurons into the striatum and substantia nigra of PD model rats, respectively. The results showed that the differentiated cells were superior to the hUC-MSCs group in terms of behavioral recovery and dopamine level improvement. It is speculated that dopaminergic neurons not only exert neuroprotective functions by secreting BDNF and other factors, but also possess cell replacement capabilities, directly replenishing dopamine in the brain.53 Existing studies still display differences between animal models and human disease pathology, and data on long-term survival and tumorigenicity are insufficient. Future work needs to be conducted in models that more closely resemble clinical pathology and to perform systematic safety evaluations. Comparison of the Three Cellular Mechanisms Against PD The MSCs derived from the three tissues display distinct emphases in PD models. BM-MSCs have the most systematic profile: after intravenous or intracerebral delivery the cells home to the substantia nigra, differentiate into TH-positive neurons, up-regulate TH and VMAT-2, enhance antioxidant-enzyme activity and down-regulate IL-6 and TNF-α, thereby improving dopamine synthesis, oxidative stress and the inflammatory milieu. In addition, their paracrine action lowers peripheral pro-inflammatory cytokines and elevates BDNF, generating a central-plus-peripheral dual modulation. Compared with BM-MSCs, AD-MSCs core advantage lies in correcting peripheral immune imbalance. They reshape the skewed Th17/Treg axis in PD by suppressing RORγt and promoting IL-10 secretion, thus blocking the aberrant immune response that triggers neuro-inflammation. Simultaneously, they secrete PTX3 to exert neuro-protection by blocking the extrinsic apoptotic pathway. In contrast, hUC-MSCs tend to favor in situ structural repair. Their primitive status endows them with robust paracrine capacity and multi-lineage potential. Numerous studies have focused on efficiently inducing them into functional, electro-physiologically active dopamine-secreting neurons. Such pre-differentiated cells have been proven superior to undifferentiated counterparts in behavioural recovery and dopamine-level elevation, achieving “on-site precise replacement” and offering a direct means to reconstruct the injured nigro-striatal circuit. At the clinical translation level, BM-MSCs possess the most clinical data. Both intravenous and intracerebral transplantation have demonstrated short-term safety, but their efficacy in treating late-stage PD is limited, and the cell quality is restricted by the age of the donor. AD-MSCs are easily obtainable, and animal experiments have shown that their protective effects are superior to conventional drugs. However, clinical trial data are scarce and the safety of their therapeutic efficacy remains to be verified. hUC-MSCs can be induced into functional dopaminergic neurons and serve as candidates for late-stage replacement therapy. However, the stability of the induced differentiation system and the insufficient survival rate after transplantation pose challenges to its clinical translation. Taken together, the three cells correspond to different pathological stages of PD. For early inflammation-driven stages, BM-MSCs or AD-MSCs may be preferred. Whereas for late stages with massive dopaminergic neuron loss, pre-differentiated hUC-MSCs may be considered. Future studies should determine-in chronic models bearing α-synuclein deposition-which pathological stage is more suitable for neural replacement and which for paracrine modulation, and then clarify the optimal indication population and delivery mode through phase-II randomised controlled trials. Therapy of Amyotrophic Lateral Sclerosis Therapy of Bone-Marrow-Derived MSCs Against ALS BM-MSC therapy for ALS remains in the early stage of clinical translation. In a clinical trial, Shamsaei G et al discovered that when ALS patients were injected with autologous BM-MSCs, it could temporarily slow down the decline rate of forced vital capacity and slow down the progression of the disease. Therefore, it is speculated that it may regulate neuroinflammation by reducing monocyte chemoattractant protein-1 (MCP-1) and other factors, inhibit glutamate excitotoxicity, and improve the nutritional conditions of the nerves.54 A phase-II trial of repeated intrathecal injection conducted by Oh et al demonstrated good safety and alleviated the short-term decline of ALSFRS-R scores. Cerebrospinal-fluid analyses revealed elevated anti-inflammatory factors and decreased pro-inflammatory factors, suggesting that BM-MSCs exert therapeutic effects by modulating the central immune micro-environment.55 These trials are, however, limited by small sample sizes and have not proven long-term clinical benefit. Basic work by Chio et al performed long-term in-vitro expansion of BM-MSCs derived from ALS patients and found that proliferative, differentiative and neurotrophic-factor-secreting capacities all declined with increasing passage number. The authors therefore advocate using early-passage MSCs for transplantation to guarantee effective anti-inflammatory and neuro-protective actions.56 Proteomic analysis of cerebrospinal fluid from BM-MSC-treated ALS patients by Reis et al identified 220 differentially expressed proteins 30 days after therapy. Protein-interaction-network and biological-enrichment analyses revealed significant clustering of extracellular matrix adhesion molecule pathways; nine proteins, including APOA1, APOE, APP and C4A, occupied core positions in the network and may represent key molecules through which MSCs remodel the extracellular environment, providing candidate targets for subsequent mechanistic studies.57 Therapy of Adipose-Derived MSCs Against ALS AD-MSCs possess potential therapeutic capacity for ALS. The Chiarotto team found that in early-stage ALS mice AD-MSCs rescued 96% of motor neurons, preserved synaptic-marker expression, suppressed glial activation, decreased pro-inflammatory factors, up-regulated HGF expression, delayed ALS progression and prolonged survival.58 Another study systemically injected AD-MSCs into SOD1 mutant mice during the clinical onset of disease, and also observed delayed deterioration of motor function. After 100 days of treatment, the number of motor neurons in the lumbar spinal cord of mice was significantly higher than that of the control group. Studies have shown that AD-MSCs significantly increased the levels of GDND and bFGF in spinal cord tissue, suggesting that they may regulate the secretion profile of glial cells through paracrine action, causing them to transform into a neuroprotective phenotype and exerting indirect neuroprotective effects.59 Although AD-MSCs retard ALS progression in multiple acute-injury or immune-imbalance animal models, clinical translation faces an efficacy gap between animals and humans: a clinical trial of single intravenous autologous AD-MSC infusion in ALS patients demonstrated safety and feasibility but failed to slow motor-function decline, illustrating that animal models cannot reproduce the progressive motor-cortex degeneration of human ALS and that paracrine signals cannot long-term maintain distal axons and neuromuscular junctions.60 More rigorous clinical trials are therefore needed to verify efficacy. Therapy of Human Umbilical Cord-Derived MSCs Against ALS The potential efficacy of hUC-MSCs against ALS has gained preliminary support in SOD1-mutant mice. After intravenous hUC-MSC administration, abnormal microglial activation in the spinal cord was suppressed, astrocytic proliferation decreased, and the expression of pro-inflammatory factors and iNOS was lowered, interrupting the vicious cycle of neuro-inflammation. In-vitro experiments further confirmed that hUC-MSCs directly inhibit the inflammatory response of mutant-SOD1 microglia and prevent their induction of astrocytes toward the toxic A1 phenotype.61 Subsequent work added that hUC-MSCs attenuate SOD1G93A-induced mitochondrial membrane potential collapse and oxidative stress and activate autophagy, relieving motor-neuron apoptosis.62 However, the efficacy may depend on the route of administration. Sironi F et al injected hUC-MSCs into the cerebral ventricles of ALS model mice and found that the cells failed to migrate to the spinal cord, activated only local p-Akt survival signalling via paracrine action, slightly delayed motor-neuron loss, but could not prevent peripheral neuromuscular-junction degeneration, so overall disease course was not significantly delayed.63 Based on pre-clinical studies, clinical trials of hUC-MSCs for ALS have been initiated. A phase-I safety trial by Barczewska’s team in 43 patients confirmed good tolerability with no serious adverse events.64 A subsequent retrospective case-control study in 67 treated patients matched with PRO-ACT database controls demonstrated prolonged median survival and reduced mortality risk.65 Li et al pointed out, however, that stem-cell efficacy is short-lived and disease continues to progress during long-term follow-up, indicating limited long-term benefit.66 Future strategies should explore repeated intrathecal injection, combined administration of neurotrophic factors or existing drugs to prolong clinical efficacy of hUC-MSCs. Dissection of Limitations of the Three MSC Therapies in ALS All three tissue-derived MSCs display limitations in ALS clinical application. The greatest obstacle for BM-MSCs is replicative senescence: during in-vitro expansion BM-MSCs from ALS patients show declining proliferation, differentiation and secretion with increasing passage, forcing cell preparation to be restricted to early passages and complicating standardised manufacture. Moreover, phase-II trials only observed short-term slowing of ALSFRS-R decline, and evidence for long-term clinical benefit remains absent. The efficacy seen with AD-MSCs in animal models virtually disappeared in single intravenous autologous clinical trials, indicating an essential gap between animal models and human pathology and necessitating slower-progression models that more closely mimic human disease. Limitations of hUC-MSCs are their short duration and restricted migration. Long-term follow-up shows continued disease progression and waning therapeutic effects over time. Intraventricularly injected hUC-MSCs fail to migrate to spinal-cord lesions, sustain only local p-Akt signalling via paracrine action, and cannot prevent continuous degeneration of distal axons and neuromuscular junctions. Figure 4 summarizes the safety, efficacy, and limitations of three types of mesenchymal stem cells (BM-MSCs, AD-MSCs, and hUC-MSCs) as observed in clinical trials for ALS. Overview of MSC therapy outcomes in ALS. BM-MSCs: safe, with transient slowing of ALSFRS-R and FVC decline in phase I/II trials. AD-MSCs: safe, but failed to slow disease progression. hUC-MSCs: safe, with early functional improvement in retrospective/prospective studies, though long-term decline persists. Phase III evidence remains pending for all sources. Created with BioRender.com. Therefore, future work should establish GMP-compliant, reproducible batch protocols and develop repeated intrathecal or intravenous regimens combined with neurotrophic factors or existing drugs to prolong clinical benefit of MSCs in ALS patients. MSCs-Derived Extracellular Vesicles: Targeted Regulatory Mechanisms The whole-cell MSCs therapies described above constitute an important route for treating neurodegenerative diseases. Their advantage lies in delivering an intact therapeutic unit that not only achieves structural replacement and neural repair through multi-lineage differentiation potential, but also responds to micro-environmental changes at lesion sites, spontaneously secretes multiple neurotrophic and anti-inflammatory factors, and exerts immunomodulation via direct cell-cell contact. Consequently, however, clinical translation is challenged by low survival, unstable homing, potential tumorigenic risk and immune rejection. In recent years investigators have shifted the breakthrough point from the stem cells themselves to extracellular vesicles-lipid-bilayer membrane vesicles released by cells. They cross the blood-brain barrier efficiently, and the proteins and nucleic acids they carry are the key molecules underlying therapeutic effects.67 Compared with living cells, EVs are highly stable, express virtually no MHC-II and carry a low risk of immune rejection. They retain the paracrine activity of the parent cells while avoiding hazards such as tumorigenicity and unstable homing. Moreover, some studies report that their therapeutic effect in neurodegenerative diseases is superior to that of whole MSCs.68 MSC-EVs are therefore regarded as the core vehicle of “cell-free therapy”, and their pre-clinical validation in AD, PD and ALS models has advanced rapidly in recent years. Biological Characteristics of EVs and Compositional Heterogeneity of MSC-EVs from Different Sources MSC-EVs act as inter-cellular communication mediators: cargo is wrapped by a lipid bilayer and targeted delivery is achieved via membrane surface proteins. MSC-EVs from different tissues share conserved proteins, yet also display unique surface molecules that provide a basis for therapeutic specificity.69 For example, human adipose MSC-EVs specifically carry the Aβ-degrading enzyme neprilysin,70 whereas human umbilical-cord MSC-EVs are enriched in synaptic-vesicle-recycling proteins such as AP2A1 and AP2B1.69 Bone-marrow MSC-EVs highly express the anti-inflammatory protein TNF-stimulated gene 6 (TSG-6) and deliver microRNA-29c-3p that targets β-secretase.71,72 Such compositional differences offer candidate molecules for disease-specific therapy. Multi-Level Mechanisms of MSC-EVs in Neurodegenerative Diseases MSC-EVs do not target a single signalling pathway; they cooperate at molecular, cellular and systemic levels to intervene at multiple points in the pathological networks of AD and PD. First, they can directly clear pathogenic proteins or modulate their metabolism. In AD models AD-MSC-EVs carry NEP that specifically degrades Aβ, lowering Aβ40 and Aβ42 levels.70 BM-MSC-EVs may, via microRNA-29c-3p or other components, inhibit β-secretase translation and expression, thereby reducing Aβ generation at the source.72,73 In PD models the TSG-6 protein enriched in BM-MSC-EVs suppresses the STAT3/miR-7/NEDD4/LRRK2 axis: TSG-6 inhibits STAT3-mediated transcriptional activation of microRNA-7 (miR-7), leading to reduced miR-7 expression, relief of its suppression of the E3 ligase NEDD4, and subsequent ubiquitination and degradation of the pathogenic protein LRRK2, ultimately decreasing α-synuclein accumulation.71 EVs exert consistent anti-inflammatory effects in AD, PD and ALS models. AD-MSC-EVs inhibit microglial polarisation toward the pro-inflammatory M1 phenotype, down-regulate activation markers CD11b and iNOS, promote anti-inflammatory mediators IL-10 and tissue inhibitor of metalloproteinases-1 (TIMP-1) and up-regulate the neurotrophic factor BDNF,74 while lowering the astrocytic activation marker glial fibrillary acidic protein (GFAP).75 AD-MSC-EVs loaded with miR-22 further suppress the NLRP3 inflammasome, reduce cleavage and activation of caspase-1, and inhibit cleavage of its substrate gasdermin D (GSDMD) into the pyroptosis-executing fragment p30-GSDMD, thereby preventing neuro-inflammatory damage.76 hUC-MSC-EVs likewise inhibit NLRP3 inflammasome activation and decrease release of IL-1β and IL-18, preventing pyroptosis-like morphological changes in microglia.77 EVs promote neurogenesis and synaptic repair, ameliorating cognitive and motor deficits. In AD models both AD- and BM-MSC-EVs enhance hippocampal neurogenesis and increase the number of newborn neurons, rescuing memory and spatial-learning deficits in APP/PS1 transgenic mice; the benefit is linked to elevated BDNF and improved synaptic plasticity.78,79 Proteomic analyses further show that hUC-MSC-EVs are enriched in adaptor-protein complex AP-2 subunits that accelerate synaptic-vesicle endocytosis and recycling, restoring neurotransmitter-release efficiency.69 In PD models hUC-MSC-EVs raise striatal dopamine and its metabolite levels and up-regulate dopaminergic-neuronal markers and neuron related proteins, improving motor function.77 Additionally, EVs modulate autophagy and inhibit cell death in a context-dependent manner. In PD models hUC-MSC-EVs up-regulate the nuclear-matrix protein special AT-rich sequence-binding protein 1 (SATB1), activate the Wnt/β-catenin pathway and inhibit excessive autophagy, as reflected by decreased LC3-II/LC3-I ratio and p62 accumulation, thereby protecting dopaminergic neurons.80 Conversely, under other experimental conditions hUC-MSC-EVs increase LC3B-II/I ratio and Beclin-1 and decrease p62, inducing protective autophagy that promotes survival of 6-OHDA-injured neuronal cells.81 The therapeutic mechanisms by which MSC-derived EVs function in neurodegenerative diseases were illustrated by Figure 5. MSC-EVs from three sources carry distinct cargoes: BM-EVs (TSG-6, miR-29c-3p), AD-EVs (neprilysin, miR-22), UC-EVs (AP2A1, AP2B1). Despite these differences, they cover four main actions in neurodegenerative disease: clearance of pathogenic proteins, suppression of microglial inflammation, enhancement of neurogenesis and synaptic repair, and modulation of autophagy to promote neuronal survival. Created with BioRender.com. Realisation of these mechanisms relies on the efficient in-vivo transport and targeted delivery of EVs. Multiple studies confirm that after intravenous or intranasal administration MSC-EVs not only effectively cross the blood-brain barrier but also specifically accumulate in diseased brain regions such as the hippocampus, neocortex and substantia nigra, where they are taken up by neurons and glia, providing a pharmacokinetic basis for precise intervention.76,77,82 Clinical Progress of EVs Therapy At present, clinical studies of EVs for neurodegenerative diseases are still in their infancy; evidence for AD and PD is limited, but trials in ALS have shown good safety and preliminary efficacy. A case report described a late-stage ALS patient whose muscle strength and functional scores improved after BM-MSC-EVs treatment without adverse reactions.83 A 2024 prospective pilot safety study enrolled ten ALS patients and further confirmed that intravenous infusion of BM-MSC-EVs is safe and stabilised disease progression in 30% of patients within three months, suggesting the ability to delay deterioration.84 These data provide a feasible basis for subsequent large-scale randomised controlled studies. Limitations and Prospects of EVs Therapy Although EV therapy is safer and easier to store than whole-cell MSCs, it still has limitations and clinical translation faces major challenges. First, unified isolation standards and dose-conversion methods are lacking. Second, their complex therapeutic mechanisms need further elucidation to enhance targeting, and large-scale clinical trials in AD and PD are still absent. Finally, one study reported that BM-MSC-EVs injection actually impaired spatial memory in healthy rats, possibly disturbing neurophysiological balance, and advised against using exosomes as memory enhancers in healthy populations.85 It is worth noting that a key but overlooked bottleneck in current extracellular vesicle research is undefined inter batch heterogeneity. Due to variable cell culture conditions, cell passage aging, and isolation protocols, the cargo composition and therapeutic consistency of extracellular vesicles are greatly altered, severely hindering stable clinical translation and reproducible neuroprotective efficacy. To promote the clinical translation of MSC-derived EVs, future research must, on the one hand, design and conduct large-scale randomised controlled clinical trials to establish safe doses and clinical efficacy. On the other hand, EVs can be functionally optimised by preconditioning, genetic engineering or chemical modification to load specific molecules and enhance targeting and therapeutic effect. This direction is a current hotspot and will be discussed in detail below. Modification and Pre-Conditioning: Impact on Stem Cell Therapy Genetic Engineering to Optimise Stem Cell Efficacy As outlined above, therapeutic potency of naive stem cells is often insufficient within complex disease milieus. Genetic or bio-engineering modifications can enhance targeting, survival and secretion, thereby improving outcomes in neurodegenerative diseases. According to the nature of the modification they can be classified as follows (Figure 6). Strategies to augment MSC therapeutic potency through genetic/bio-engineering modification (such as neurotrophic factors, functional enzymes or proteins, transcription factors (TFs), or non-coding RNAs (miRNAs)) and ex vivo preconditioning (such as antioxidants, drugs, or cytokine cocktails). These enhancements aim to improve MSC homing, survival, paracrine output, synaptic remodeling, and toxic-protein clearance while mitigating immunogenic and tumorigenic risks. The downward arrow indicates a decrease in immunogenicity. The cross indicates inhibition of tumor formation. The circle-slash symbol represents the removal of toxic proteins. Created with BioRender.com. Neurotrophic-Factor Gene Modification Introduction of exogenous neurotrophic-factor genes enables stem cells to secrete neurotrophic factors continuously, exerting multiple effects: protecting host neurons, improving the local micro-environment, and promoting survival and differentiation of the graft itself. Owing to easy access and low immunogenicity, BM-MSCs are the most frequently used carrier cells. Multiple neurotrophic-factor modifications have proved effective, yet efficacy varies with models. In the 6-OHDA neurotoxin model BM-MSCs over-expressing BDNF or nerve growth factor (β-NGF) both out-perform controls. A representative study by Zhang stably expressed BDNF in BM-MSCs and transplanted them via the lateral ventricle; cells migrated with cerebrospinal fluid and markedly up-regulated TH protein in the injured side. BDNF presumably activates tropomyosin receptor kinase B (TrkB) to initiate anti-apoptotic and anti-oxidant downstream signals, protecting neurons and improving rat behavior.86 Similar work grafting β-NGF-modified BM-MSCs directly into the striatum produced clearly superior behavioural recovery. Histology confirmed long-term survival, migration and integration of grafted cells with sustained local β-NGF expression, indicating that neurotrophic protection and MSC-mediated cell replacement may act synergistically for better outcome.87 However, the achievements made in the classical neurotoxin model face challenges in the inflammation-driven model. Hoban et al used an LPS-injected substantia-nigra PD model. GDNF-over-expressing BM-MSCs protected TH-positive fibres locally, but failed to improve motor function, reduce nigral inflammation or prevent dopaminergic neuron loss, mainly attributed to low graft survival within the inflammatory milieu.88 Thus, raising survival of transplanted cells in the host brain may be a prerequisite for therapeutic efficacy when PD pathology is inflammation-driven. Besides BM-MSCs, AD-MSCs also show potential as gene-engineering vehicles. GDNF-over-expressing human AD-MSCs significantly improve post-graft survival and enhance neural-lineage differentiation via paracrine action. In PD mice this self-reinforcing effect translates into stronger host-neural protection, reflected by increased density of TH- and NeuN-positive cells in the nigro-striatal pathway and significant improvement in motor function, demonstrating that neurotrophic-factor modification protects host neurons and simultaneously modulates graft function to enhance efficacy.89 Given the strong multi-lineage potential of hUC-MSCs, neurotrophic-factor gene modification can be combined with directed differentiation. A recent study first induced the phenotypic differentiation of hUC-MSCs towards cholinergic neurons, followed by gene modification with brain-derived neurotrophic factor (BDNF). After being transplanted into AD rats, these cells continuously released BDNF, and the release of acetylcholine and the level of choline acetyltransferase increased simultaneously. This was accompanied by the activation of glial cells, a reduction in Aβ deposition, a decrease in apoptotic neurons, and an increase in new neurons, thereby significantly improving spatial learning and memory abilities.90 Functional-Enzyme or Protein Gene Modification Delivery of genes encoding key enzymes for neurotransmitter synthesis or pathogenic-protein degradation, or regulators of disease-related proteins, allows stem cells to intervene directly in pathogenesis. The core pathology of PD is severe striatal dopamine depletion. Genetic engineering can force stem cells to express key enzymes of dopamine synthesis to replenish the neurotransmitter. Research has found that after transplantation of TH and Neuturin (NTN) dual-gene co-modified BM-MSCs into a PD rat model, the local DA and its metabolite DOPAC levels in the striatum increased by approximately 100%, and the expression levels of TH and NTN proteins also increased. At the same time, it effectively downregulated the density of postsynaptic dopamine D2 receptors. The behavioral deficits of the rats were also effectively improved.91 This indicates that this strategy has a synergistic effect in promoting DA synthesis and protecting dopamine neurons, and provides preliminary evidence for multi-gene combination therapy. The strategy is, however, limited: DA biosynthesis requires coordinated action of TH, DDC and GCH1. TH alone exhibits poor catalytic efficiency owing to dependence on endogenous cofactor BH4 and DDC activity. Recent studies have built complex multi-enzyme co-expression systems with stronger synthetic capacity. Wang et al simultaneously introduced TH, DDC and GCH1 into BM-MSCs to create DA-BM-MSCs capable of autonomous high-efficiency DA synthesis.92 The concentration of DA in the supernatant of its in vitro culture reached up to 4128 ng/mL, which was 12.4 times higher than that of the control group, and the production was far beyond that of cells overexpressing only TH or both genes. It could continuously and stably provide physiological levels of DA in vivo. Secondly, DA-BM-MSCs exhibited a more comprehensive role in neural functional recovery, improving the complex motor coordination ability of model animals. The heat map of animal activity trajectories showed that their behavioral patterns were close to those of the control group. Moreover, DA-BM-MSCs displayed stronger survival, differentiation and integration capacity. Grafted cells survived long-term, differentiated into TH-positive dopaminergic neurons, GABA-ergic neurons, glutamatergic neurons and astrocytes, and formed synaptic connections with host neurons. Finally, they significantly reduced striatal pro-inflammatory factors, inhibited microglial activation and ameliorated neuro-inflammation, demonstrating multi-target synergy and offering a new strategy for genetically engineered cell therapy in PD. Another study employed hUC-MSCs modified with the NEP gene to treat an AD mouse model. The results indicated that both NEP-modified hUC-MSCs and control hUC-MSCs improved cognitive function in the model mice, reduced Aβ plaque deposition in the brain, and suppressed neuroinflammation. However, EVs released from the NEP-overexpressing group contained higher levels of NEP protein, thereby exerting stronger neuroregenerative, neurotrophic support, and anti-inflammatory effects in the hippocampus.93 Besides introducing neurotransmitter-synthetic enzymes, modification with extracellular-matrix and micro-environment-regulatory proteins has become a novel strategy for restoring neural function. Feng et al constructed TIMP2 gene-modified hUC-MSCs targeting TIMP2, an important extracellular matrix regulatory protein. The secreted TIMP2 protein can cross the blood-brain barrier, activate the expression of oncogene homolog (c-Fos), and enhance synaptic plasticity. At the same time, it can maintain the homeostasis of the neuronal microenvironment, enhance cell adhesion and signal transduction. The modified hUC-MSCs can also activate endogenous anti-apoptosis and cell proliferation pathways, improving the learning and memory abilities of AD animal models through the above effects.94 Transcription-Factor Gene Modification Two bottlenecks in stem-cell therapy are replicative senescence during in-vitro expansion and low post-graft survival. Over-expressing specific transcription factors can regulate stem-cell survival and homing and enhance function to optimise therapeutic efficacy. Zhang et al focused on the transcription factor forkhead box Q1 (FOXQ1) in hUC-MSCs.95 FOXQ1 can activate the transcription of SIRT1, downregulate the expression of aging-related proteins such as p16, thereby delaying the aging of hUC-MSCs and enhancing their proliferation and migration abilities. In AD model mice, hUC-MSCs overexpressing FOXQ1 exhibited stronger intracerebral migration and survival abilities. hUC-MSCs not only reduced the expression of aging-related proteins in the host hippocampus, but also decreased the levels of phosphorylated Tau, enhanced antioxidant enzyme activity, and improved cognitive impairment. Wang et al explored the nuclear-receptor-related factor Nurr1 in BM-MSC therapy for PD. After transplanting Nurr1 gene-modified BM-MSCs into the striatum of 6-OHDA-injured rats, the survival and migration abilities of BM-MSCs were enhanced, and the expression of TH in the substantia nigra and striatum of the host brain was significantly upregulated, thereby improving the rotational behavior of the model rats.96 Non-Coding-RNA Engineering Genetically up- or down-regulating specific non-coding RNAs in stem cells is another strategy to enhance efficacy. Altering the stem-cell secretome-especially the RNA cargo of secreted exosomes-can modulate gene-expression networks in target cells and intervene in disease pathways. In AD therapy, specific RNA inhibition can raise Aβ clearance. Chen et al found that the expression of Aβ-degrading enzyme NEP is inhibited during the progression of AD. Through bioinformatics screening, they confirmed that miR-211-5p is a negative regulator of NEP. Based on this, they constructed hUC-MSCs that overexpressed miR-211-5p and isolated their exosomes. In SH-SY5Y cells damaged by Aβ1-40, this engineered exosome was able to effectively upregulate the expression level of NEP within the cells, thereby significantly reducing Aβ-induced cell apoptosis, decreased migration ability, and secretion of pro-inflammatory factors. However, when NEP was knocked down, the protective effect was completely counteracted.97 For PD, an miRNA over-expression approach was adopted. Li et al found that the expression of miR-188-3p was decreased in the serum of PD patients. Bioinformatics analysis predicted and verified that its target genes were Cyclin-Dependent Kinase 5 (CDK5) and NLRP3. Exosomes secreted by miR-188-3p-over-expressing AD-MSCs were used to treat MPTP-induced PD mice. The exosomes were effectively delivered to lesions, significantly suppressed CDK5 and NLRP3 expression in the substantia nigra, thereby inhibiting excessive autophagy and pyroptosis and increasing TH-positive neuron survival. In MPP+-treated MN9D cells the same treatment promoted proliferation and inhibited apoptosis of injured dopaminergic cells. The study demonstrates that single miRNA over-expression can enrich functional sequences in exosomes; after delivery to lesions the sequences intervene in autophagy dysregulation and neuro-inflammation in PD, achieving synergistic protection.98 Limitations and Prospects of Genetically Engineered Stem Cells Although genetic or bio-engineering modifications can improve stem-cell survival, targeting and therapeutic outcome in complex pathological environments, translation from bench to bedside still faces risks and challenges. The clinical applications of MSCs from different sources have shown varying levels of safety and efficacy, as summarized in Table 1. Potential tumorigenic risks of viral vectors and over-expression of transgenes must be evaluated. Moreover, stem-cell engraftment in lesions remains low. Future work can combine novel non-viral vectors with stem-cell surface engineering and gene modification to enhance both safety and homing/survival capacity. In order to enhance the targeting ability of MSCs, Zhang et al labeled hUC-MSCs with superparamagnetic nanoparticles containing Fe3O4 and polydopamine (Fe3O4@PDA). Fe3O4@PDA has improved the efficiency of hUC-MSCs entering the brain. In addition, APP/PS1 transgenic mice treated with Fe3O4@PDA labeled hUC-MSCs showed significant improvement in cognitive ability compared to mice treated with only hUC-MSCs.99 Table 1Clinical Trials of MSCs Therapy in Neurodegenerative DiseasesStudy TypeTrial Design (Phase)Patients (Age Range)MSC SourceDoseSafetyEfficacyRef.Controlled TrialRandomized, double-blind, placebo-controlled (Phase 1)33 mild AD (mean 71.2 years)Allogeneic BM-MSCLow: 2×106 cells; High: 1×108 cellsNo treatment-related SAEs; no ARIA; primary safety endpoint metLow-dose improved cognitive and functional scores; high-dose increased biomarkers and hippocampal volume but no clinical benefit[100]Controlled TrialRandomized, double-blind, placebo-controlled (Phase 2a)49 mild AD (mean age 74.1 years)Allogeneic BM-MSC (laromestrocel)Single 25×106 cells; repeated 25×106 cells ×4 monthlyNo infusion reactions, no ARIAs, no hypersensitivity. No deathsActive groups improved on cognitive and functional measures; brain atrophy and neuroinflammation reduced[25]Controlled TrialRandomized, double-blind, placebo-controlled (Phase 2)45 mild-to-moderate PD (mean age 64–69 years)Allogeneic BM-MSCs10×106 cells/kg;3 infusions (weeks 1, 18, 36)10 mild transient AEs; no severe AEs; no immunogenicityThree-infusion group improved vs placebo. MDS-UPDRS total improved[101]Controlled TrialRandomized, double-blind, placebo-controlled (Phase 2)24 patients with early-to-moderate PD (mean age 60.6 years)Autologous AD-MSCs2×108 cells per infusion; 6 infusions at Weeks 0, 4, 8, 16, 24, and 32Well tolerated. No deaths or clinically meaningful lab changesNo significant difference vs placebo for primary or secondary endpoints. Trial not powered for definitive efficacy[102]Controlled TrialObservational study with external control (Phase 2/3)157 ALS patients with symptom duration <2 yearsAutologous BM-MSCSingle-cycle: two injections 1 month apartMost AEs mild/transient; SAEs 26.85% (mainly disease-related)Significant survival benefit vs external control; median survival not reached in BM-MSC group vs 24 months in control[103]Controlled TrialRandomized, double-blind (Phase 1/2)40 ALS patients (age ≥18 years)Autologous AD-MSCs1×106, 2×106, or 4×106 cells/kgFeasible and safe; procedural phlebitis/thrombosis reportedNo efficacy vs placebo; ALSFRS-R declined as expected; no dose-response or survival benefit[60]Pilot StudyOpen-label, single-center, no placebo control (Phase 1)20 PD (mean 66 years, range 45–78)Allogeneic BM-MSC1×106, 3×106, 6×106, or 10×106 cells/kgNo infusion-related SAEs or antibody responses. Most common AEs: dyskinesia and hypertension (20% each)Highest dose showed sustained motor improvement at 52 weeks; biomarkers improved. Trial not powered for efficacy[39]Pilot StudyOpen-label, single-arm, safety-focused(Phase 1)43 ALS patients (mean age 57.3 years)Allogeneic WJ-MSC (umbilical cord-derived)0.42×106 cells/kg (range 0.27–0.7); 1–3 injections at 2-month intervalsNo serious AEs; one transient headache resolved spontaneously. Procedure well toleratedNot evaluated (safety-focused study)[64]Case ReportCase report (single patient, no control)1 patient (age 73), 3-year history of ADAutologous BM-MSCsNot specifiedNo adverse events reportedImprovement observed on day 1, lasting several weeks before gradual regression[104]Case ReportSingle case report, no control group1 male ALS patient (age 46)Autologous AD-MSCsSix infusions (3.8–5.9×107 per infusion)No adverse reactions observed during or after any infusionPatient survived >10 years without tube feeding; ALSFRS-R improved from 31 to 46 but fluctuated; no control group, cannot attribute to treatment[105]Abbreviations: WJ-MSC, wharton’s jelly-derived mesenchymal stem cell; AE, adverse event; SAE, serious adverse event; ARIA, amyloid related imaging abnormalities; ALSFRS-R, amyotrophic lateral sclerosis functional rating scale-revised; MDS-UPDRS, movement disorder society-unified parkinson’s disease rating scale. Clinical Trials of MSCs Therapy in Neurodegenerative Diseases Abbreviations: WJ-MSC, wharton’s jelly-derived mesenchymal stem cell; AE, adverse event; SAE, serious adverse event; ARIA, amyloid related imaging abnormalities; ALSFRS-R, amyotrophic lateral sclerosis functional rating scale-revised; MDS-UPDRS, movement disorder society-unified parkinson’s disease rating scale. Pre-Conditioning to Optimise Stem-Cell Therapeutic Efficacy Compared with genetic engineering methods, pre-treatment of stem cells in vitro is a simpler and more effective method for enhancing efficacy. Specific physical, chemical or biological factor pre-treatment can significantly enhance stem-cell survival, migration, secretion and exosome targeting capacity. Multiple pre-conditioning agents may reinforce stem-cell function through the same molecular pathway. Melatonin and deferoxamine converge on the HIF-1α-Nrf2 axis: melatonin inhibits synthesis of prolyl hydroxylase domain-containing protein 2 (PHD2) via melatonin receptors 1 and 2 (MT1/2), whereas deferoxamine chelates Fe2+ and blocks PHD2 activity, stabilising hypoxia-inducible factor-1α (HIF-1α) and permitting its nuclear entry.106–108 Nuclear HIF-1α increases transcription of VEGF, erythropoietin (EPO) and glucose transporter GLUT1, and sequesters Kelch-like ECH-associated protein 1 (Keap1), releasing Nrf2. Nrf2 then initiates HO-1 and other antioxidant genes and suppresses pro-apoptotic proteins. As a result, MSCs maintain energy supply under hypoxia and oxidative stress, raise survival rate and continuously express neural-lineage transcription factors. Secreted VEGF and EPO promote angiogenesis, improve perfusion and nutrient delivery and favour long-term post-graft survival, anti-oxidation and neural repair. Facing intense oxidative damage in AD pathology, more potent antioxidant pre-conditioning agents have been tested. Researchers pre-treated stem cells with berberine and dimethyl fumarate; both agents, independent of upstream signals, modify the Nrf2 inhibitor Keap1, rapidly driving Nrf2 nuclear translocation and powerfully initiating downstream antioxidant-gene transcription. The induced antioxidant-enzyme expression is faster and more pronounced, effectively neutralising acute oxidative stress. Moreover, strong Nrf2 signaling can effectively regulate excessive cell apoptosis. The established redox homeostasis can enable MSCs to evolve from a simple survival state to enhance their repair ability in pathological environments.109,110 In addition, after pre-treating stem cells with a combination of cytokine complexes, the targeting ability of the extracellular vesicles secreted by these cells was enhanced, and they could simultaneously target multiple core pathologies of AD. For example, they specifically upregulated the expression of Aβ degradation enzymes such as neprilysin, transformed microglia from an inflammatory phenotype to an phagocytic phenotype, and activated the Nrf2 antioxidant pathway, etc.111 To some extent this optimisation overcomes the above-mentioned limitations in targeting and efficacy of MSC exosomes, offering a possibility for clinical translation of stem-cell-EV therapy. Conclusions and Perspectives Mesenchymal stem-cell and stem-cell-derived extracellular-vesicle therapies have opened a new avenue for treating neurodegenerative diseases. The present review systematically demonstrates that MSCs from different tissues can exert neuro-protection via paracrine action, immune modulation and protein clearance, yet each source displays distinct advantages and drawbacks in mechanism and clinical application. Extracellular vesicles refine the therapeutic functions of stem cells, while the introduction of genetic engineering and pre-conditioning strategies further enhances targeting and efficacy. However, there are still many limitations in the current research. It is critical to emphasize that current evidence is predominantly preclinical, with robust and conclusive clinical data still lacking to support widespread clinical translation. At the cellular level, BM-MSCs are limited by the donor’s age and are difficult to obtain; AD-MSCs have a narrow optimal dosage range, and high-dose transplantation may even lead to a decrease in efficacy; hUC-MSCs are constrained by issues such as low survival rate after transplantation and limited long-term efficacy. Mechanistically, most investigations remain at the level of phenomenological description or verification of known pathways. In recent years, although technologies such as single-cell genomics have been able to verify the source and functional heterogeneity of MSCs, there is still a lack of systematic evidence and a unified theoretical explanation for the underlying mechanisms of the therapeutic efficacy differences, the dynamic changes of paracrine effects in the body, and the specific interactions with the host microenvironment. Researchers have used single-cell sequencing to reveal the complex lineage differentiation trajectory from MSCs to bone, cartilage, and fat, providing a detailed analysis of MSC lineage differentiation.112 In addition, pre-clinical work relies mainly on transgenic or toxin-induced models that recapitulate only fragments of human pathology, leading to a translational gap between animal outcomes and clinical benefit in AD, PD and ALS. Existing clinical trials are generally small, short in follow-up and inadequately controlled, and thus fail to provide high-level evidence. Greater attention should be given to the need for large randomized controlled clinical trials to fill the clinical evidence gap and validate the true therapeutic value of MSC-based therapies. Furthermore, cell-manufacturing protocols, delivery routes, treatment timing and dosage have not been standardised. Finally, potential risks of viral vectors in genetic engineering, together with the long-term safety and tumorigenicity of stem-cell transplantation, also constrain clinical application. Regarding the unresolved issues such as unclear mechanisms, conversion barriers, and lack of standardization, future research should focus on targeted and practical optimization directions. First, it is necessary to combine multi-omics technologies with human neural organoid models to systematically analyze the core regulatory nodes of the dynamic interaction between MSC and the neural immune microenvironment. Secondly, by integrating cell pretreatment, targeted gene modification, and biocompatible biomaterials technology, an engineered MSC with pathological signal perception, on-demand release of therapeutic factors, and significantly improved in vivo survival and homing abilities will be constructed to address the core issues of insufficient therapeutic efficacy and poor targeting accuracy of primitive stem cells. Thirdly, carry out large-scale, multi center, and strictly based on biomarker stratified design clinical trials, establish precise and personalized treatment plans, and break down the barriers to the transformation of basic research results into clinical benefits. Lastly, a scalable and quality-controllable GMP platform for extracellular vesicles must be established. For MSC-EVs, in order to enhance its conversion value, future research should prioritize the standardization of MSC-EVs efficacy determination and the development of systematic administration routes and regimens. The MSC EVs obtained by different preparation processes currently have significant differences in particle size distribution, protein and nucleic acid components, purity, and recovery rate, and lack a unified definition standard, which directly leads to poor experimental reproducibility. In the future, it is necessary to unify the separation, purification, enrichment, and quality control processes of MSC EVs, and establish a universal EV characteristic identification standard.113 Secondly, regarding the efficacy evaluation of MSC EV, it is necessary to establish in vitro efficacy testing models and quantitative evaluation indicators in the future, clarify the effective efficacy thresholds corresponding to different indications, and provide data support for clinical dose setting and formulation quality control.114,115 Finally, there is currently a lack of systematic research on the administration of MSC EVs. The tissue distribution, targeted enrichment efficiency, and bioavailability of different administration routes are not yet clear, and there is no unified standard for administration frequency, course of treatment, and dosage. In the future, targeted delivery solutions need to be developed for different diseases to reduce non-specific consumption of EVs and increase lesion enrichment concentration. Only then will MSC therapy be transformed from an “experimental phenomenon” into a “predictable clinical tool”, and eventually it may be possible to improve the quality of life of patients with neurodegenerative diseases. Funding Statement This work was supported by the Science and Technology Development Plan Projects of Jilin Province (Grant No. 20250204072YY). Author Contributions All authors made a significant contribution to the work reported, whether that is in the conception, study design, execution, acquisition of data, analysis and interpretation, or in all these areas; took part in drafting, revising or critically reviewing the article; gave final approval of the version to be published; have agreed on the journal to which the article has been submitted; and agree to be accountable for all aspects of the work. Disclosure The authors declare no competing interests. References