Introduction

Alzheimer's disease (AD) is the most common type of dementia, the different forms of neurodegenerative condition(NDC) that results in the disruption of neural functions and cognition, leading to memory loss and eventually death(1, 2). In the United States alone it is estimated that by 2050, new cases of AD will arise every 33 seconds becoming the sixth leading cause of death in the United States and the fifth leading cause of death in Americans age over 65 years, while in China, AD is expected to be a major medical problem owing to the expected increase in aging population to 400million by 2050(3-5). Moreover, while initially believed to affect those over 65, Alzheimer's is now characterized as an early-onset progressive disease that can begin at the age of 25-30(6), affecting over 50 million people globally(7). Concurrently, with the gross improvement in healthcare access, there is an increase in the older populations necessitating thorough research into linked causes, to alleviate the burden of Alzheimer’s disease (AD) financially and psychosocially(2).

At present, 4 major pathophysiological hypotheses of AD are accepted: the Amyloid Beta (Aβ) hypothesis, the Tau protein hypothesis, the Glutamate Cytotoxicity Hypothesis, and the cholinergic hypothesis, as shown in Figure 1(A)(2, 8). In the Aβ hypothesis, the breakdown of Amyloid Precursor Protein(APP) by β- and γ-secretases (rather than α-secretases) leads to the formation of insoluble Aβ, which aggregates to form neurotoxic plaques(1, 9). In the tau protein aggregation hypothesis, normally soluble tau proteins are hyperphosphorylated, destabilizing neuronal microtubules and leading to the formation of neurofibrillary tangles, therebycausing cytoskeletal dysfunction and neurodegeneration(10). Both accumulated tau neurofibrillary tangles(NFT) and Aβ potentially cause neuroinflammation, further exacerbating neuronal damage(5, 8). Currently, there are no confirmed treatments for AD, with existing options aiming only to suppress the progression of the disease, such as by facilitating the clearance of NFT and Aβ. FDA-approved drugs for treating include Donepezil and Rivastigmine, both acetylcholinesterase inhibitors, administered orally, with the latter also acting as a butyrylcholinesterase inhibitor to reduce amyloid-beta production (2, 7, 11, 12).

Stem technology, constituting the therapeutic application of undifferentiated cells characterized by self-renewal and multi-lineage differentiation, has been a promising field of research owing to their unique ability to migrate to-sites of injury-tropic propertiesand secrete neurotrophic factors that support cellular survival(12, 13); With successes noted for diseases like leukemia, lymphoma, and certain bone marrow disorders, they have shown potential as therapeutic solutions for both acute and chronic neurological conditions, with multiple research studies being reported on their use for stroke, amyotrophic lateral sclerosis, and Parkinson's diseaseand now AD(8, 14-16)respectively. Specifically, in AD, stem cells such as Neural stem cells, Mesenchymal stem cells (MSCs), and induced Pluripotent Stem Cells (iPSCs) have been extensively investigated in preclinical models to assess their ability to clear amyloid-beta plaques, reduce tau phosphorylation, and improve cognitive function(8, 17).

Neural stem cells are self-renewing pluripotent cells capable of differentiating into the three major cell types of the central nervous system: neurons, astrocytes, and oligodendrocytes; and are typically obtained from the subventricular zone and the subgranular zone of the hippocampus in adults, or derived from exogenous sources such as embryonic tissues or reprogrammed somatic cells (iPSCs)(8, 17). As evident from the papers used in this review, current research primarilyuses mouse NSCs, which often do not translate one-to-one to human physiology due to differences in neurogenesis rates, inflammatory responses, and the complex pathological environment of the aging human brain(16). In AD mouse models, NSCs have been found to secrete soluble neurotrophic factors and cytokines that protect existing neural cells, significantly increase neuronal mitochondrial biogenesis, and enhance synaptic density, ultimately improving cognitive function(15, 18). This mechanism of ameliorating is referred to as the bystander effect, in that, unlike typical pharmaceutical therapeutics such as cholinesterase inhibitors or amyloid-targeting monoclonal antibodies that aim to delay disease progression or directly clear amyloid beta peptides, NSCs do not directly stop the accumulation of NFTs or amyloid plaques(12, 14, 19, 20).

In this review, I highlight the current reported mechanism of how NSCs modulate the neurotoxic microenvironment through paracrine signaling, and the related nuances, therebydemonstrating the progress of NSCs as potential cell therapies for AD(20, 21).

Neural Stem Cells can differentiate into the major neural cell types.

A hallmark of AD is neuronal damage, particularly axonaldegeneration, as described by tau‑protein pathophysiology. Subsequently, this is exacerbated by neuroinflammation inducedby amyloid plaque deposition(2, 8). Thus, the replacement of pathologic neuronal cells presents a compelling treatment option for AD that can be achieved only with stem cells. In a study using 10‑month‑old double‑transgenic mice, Zhang, W. et al. (2014) demonstrated that NSCs could migrate and differentiate into neurons, as well as astrocytes and oligodendrocytes, eight weeks post‑transplantation(18). The study found that a significant portion of the transplanted NSCs expressed, in vivo, the neuronal marker Tuj1 (β‑tubulin III), and the glial markers GFAP for astrocytes and GalC for oligodendrocytes; however, they also noticed that the proportion of cells differentiating into neurons was lower (13.5% of EGFP⁺ cells) compared with astrocytes (34.5%) and oligodendrocytes (32.6%)(18). Similar results (Figure 1, B(G)) were obtained by Blurton-Jones et al.(2009) using a 3xAD mouse model of the same age(12).

Although AD is mostly a localised brain pathology, the NFT and plaques result in downstream effects, such as positive feedback-mediated neuroinflammation, hence affecting all cell types in the brain(14, 21-23). Moreover, growing research indicates that glial cells may play important roles in the brain that were previously believed(23-25).Consequently, the ability to differentiate into diverse neuronal subtypes makes stem cells a strong therapeutic candidate, not only to replace neurons but also to provide modulatory effects that may otherwise be impaired in AD. For instance, microglia and astrocytes are now recognized as active regulators of synaptic pruning and remodeling, where they directly eliminate dysfunctional synapses to maintain circuit integrity. Furthermore, astrocytes play a critical role in the glymphatic clearance of Aβ and tau proteins via Aquaporin 4 channels, a process that becomes severely impaired as they transition into neurotoxic, reactive phenotypes during disease progression(25).What remainsunexamined in these studies is the NSC'stendency to differentiate into more glia than neurons. While it may appear to be an intuitive consequence, given that the brain contains a higher percentage of glia, further probing may reveal additional roles in neural cell development and the progressionof AD.

It is important to note that although we can differentiate into the various neural cells, achieving a terminally differentiated state is what we need; otherwise, there may be potential for these cells to diverge into a tumoragenic state. Two studies described here report that this is not always the case. As discussed below, Martino and Pluchino (2006)propose that stem cells have a propensity to maintain an undifferentiated state to facilitate their action by a bystander stader effect through a bimodal mechanism(14). Similarly, and unlike the positive results observed with mouse NSCs, Marsh et al. (2017) demonstrated that while HuCNS-SC human NSCs can migrate extensively within the brain, they fail to differentiate(20).The study aimed to emulate a typical clinical application of stem cells by utilizing the Rag-5xfAD immune-deficient transgenic model of AD, with HuCNS-SCs injected into the hippocampus.Unlike the majority of tests conducted with mouse NSCs, the effects of HuCNS-SCs were evaluated five months post-transplantation rather than three.They found that, following implantation, the majority of the HuCNS-SCs express immaturity markers, doublecortin markers, with none expressing neuronal NeuN, APC oligodendroglial marker, and GFAP astrocytic marker, which usually indicates terminal differentiation into mature neuronal or glial lineages(20).

Figure 1.[*Block* *A]. Summary of the pathophysiology of AD*(10). [Block B and C] NSCs ameliorate AD symptoms through the bystander effect. B(G) NSCs can differentiate to neural cell phenotypes; B(A, B, F-H) increased synaptic density, mediated by NSC BDNF. J of Neuroscience Research, Volume: 92, Issue: 2, Pages: 185-194 [Block C] (E-H, Left) Improved cognitive performance post NSC implantation despite no change in Aβ levels, (A-B, D-E, Right)(12)*. Proc. Natl. Acad. Sci. U.S.A. 106 (32) 13594-13599*

Moreover, the cells formed ectopic clusters in the lateral ventricle, which fortunately exhibited no evidence of benign activity or invasion of adjacent tissues. These findings demonstrate the need for further investigation of the differentiation timelines and states of HuCNS-SCs(and mouse NSCs) to optimize their therapeutic potential for NDCs.

Transplanted NSCs rescue cognitivedeficits without altering plaques

Using various mouse models, several studies have reported that transplanted NSCs can ameliorate cognitive deficits resulting from AD-related neuronal degeneration. This is typically assessedusing the Morris Water Maze(MWM) test and context-dependent novel object recognition, the former being a standardized test in cognitive and behavioural studies. In both tests, Blurton-Jones et al.(2009)found that, in the MWM, treated mice displayed lower escape latencies to locate the hidden platform, Figure 1,C (E-H),indicating enhanced spatial navigation and hippocampal-dependent learning(12). In the novel object recognition task, these mice demonstrated a greater preference for exploring the novel object over the familiar one, reflecting a restoration of recognition memory. Zhanget al. (2014) (12-month-old mice) and Zhang et al.(2013)show the same findings 8 weeks post-transplantation of NSCs(18, 19).

Although it has been shown that NSCs tend to migrate to regions of brain lesions such as inflamed sites (e.g., modulated human SDF-1α as reported by Imitola et al. (2004) using hNSCs), a fluorescence assay of the migration and localisation of the differentiated stem cells showed limited location within amyloid plaques, reinforced by an ELISA assay of soluble and insoluble levels of Aβ40 and Aβ42 by ELISA, which revealed no differences before and after NSC transplantationFigure 1,C (AB, DE)(12, 13). The effectiveness of the stem cells is thus explained by an indirect effect in which the stem cells modulate trophic factors such as BDNF and Synaptophysin. In the tau pathology, accumulation of NFT’s is accompanied by damage to neuronal axons, and hence neuronal synapses. Blurton-Jones et al. (2009) found that NSC transplantation led to a significant 67% increase in synaptic density within the hippocampus, correlated with an elevated expression of synaptophysin(12)Figure 1,B (F-H). In a BDNF knockdown experiment, the depletion of NSC-derived BDNF resulted in an absence of cognitive improvement and limited synaptic density restoration, establishing the NSC bystander mechanism, in which NSCs facilitate synaptic regeneration indirectly through the secretion of BDNF, influencing synaptic density markers like synaptophysin. Similar outcomes are reported byZhang et al. (2014) and Zhang et al. (2013), the formerfor amyloid plaques, showing that NSC neither reduces deposition nor reduces accumulated plaques, but with rescued long-term potentiation through the establishment of new synapses(18, 19).

Inconclusive outcomes

For Alzheimer’s disease, stem cells must be able to ameliorate the various underlying pathologies, in addition to addressing overall concerns about their use, such as their ability to differentiate appropriately and integrate into the existing neural niche, rather than forming malignant tissues. For complex conditions like AD, however, achieving consistent findings across multiple studies remains challenging and continues to limit clinical translatability. For instance, the differentiation potential of NSCs into neuronal phenotypes, as well as their therapeutic impact via bystander effects, has only been reported with near-uniform consistency across different research groups. By using a more clinically relevant model and increasing the implantation period, Marsh et al. (2017) found almost entirely contrasting results compared with all previous findings, including the limited ability of NSCs to terminally differentiate, the formation of ectopic clusters, the lack of improved cognition, and bystander effects(Figure 2.)(20. These contradictory findings may be due to variations in experimental design, such as the type of AD model, the type and source of NSCs, and the implantation period(20). Previously, Lee et al. (2015) reported that hNSCs could decrease tau phosphorylation and, to a limited extent, amyloid-β levels. The study further showed that this occurred through a bystander mechanism, in which Trk–Akt–mediated suppression of GSK3β (glycogen synthase kinase 3β) activity, a major tau-phosphorylating kinase, coupled with reduced BACE1 (β-site amyloid precursor protein cleaving enzyme 1) levels. Both processes are modulated by hNSC secretion of BDNF(5).

Figure 2.Marsh, Samuel E. et al. report contradictory outcomes using hNSCs in a 5xAD mouse model. Showing no improvement in cognitive performance (A-C), no difference in Aβ40 or Aβ42 and BDNF levels in the hippocampus after NSC implantation (D-E, and F respectively)(20).Marsh, Samuel E. et al. Stem Cell Reports, Volume 8, Issue 2, 235 – 248.

Regarding the terminally differentiated hNSCs reported byMarsh et al. (2017), however, a bimodal mechanism of the bystander effect proposed by Martino and Pluchino (2006)offers a strong conceptual framework for understanding why transplanted neural stem cells tend to persist in an undifferentiated state rather than consistently achieving full terminal differentiation(14, 20). They observed that across diverse experimental CNS disease models, functional improvement after adult NSC transplantation shows only a weak relationship to the number of graft-derived, mature neurons or oligodendrocytes. Instead, recovery more closely associates with the continued presence of cells retaining progenitor markers such as nestin, together with elevated in vivo levels of key neurotrophic factors, including nerve growth factor, brain-derived neurotrophic factor, ciliary neurotrophic factor, and glial cell line-derived neurotrophic factor, as well as cytokines(14).

This, therefore, alludes to NSCs mediating a dual bystander action: first, they secrete neurotrophic factors that support the survival, proliferation, and function of endogenous neuronal and glial progenitors while limiting glial scar formation; second, they reshape the inflammatory milieu by releasing immunomodulatory molecules and expressing chemokine receptors and adhesion molecules that influence the behavior of infiltrating and resident immune cells. Martino and Pluchino (2006) further propose that these effects are orchestrated within atypical ectopic perivascular niches, where intravenously delivered NSCs accumulate and are continuously exposed to local cues such as BMP4(Bone Morphogenic factor 4), noggin, and pro-inflammatory cytokines(14). These niche-derived signals favor long-term maintenance of an undifferentiated, therapeutically active state, thereby favoring neuroprotection and immunomodulation over extensive terminal differentiation and large-scale structural cell replacement. Owing to the different reports, it is reasonable to reiterate the statement byMarsh et al. (2017), on the reporting of both ‘good and bad’ results and data in the research on Stem cells as both are likely to benefit the progress of the field(20).

Conclusion

Neural stem cells and related stem cell populations present a promising but still immature therapeutic avenue for Alzheimer’s disease. In multiple AD mouse models, NSCs can migrate to pathological sites, generate neural and glial progeny, and improve cognition largely through a bystander mechanism, secreting neurotrophic and immunomodulatory factors such as BDNF that enhance synaptic plasticity and modulate neuroinflammation, rather than directly clearing amyloid or tau pathology.

However, translational progress is constrained by inconsistent preclinical outcomes, particularly in studies using human NSCs and longer follow-up periods, which report limited terminal differentiation, ectopic clustering, and a lack of cognitive benefit. The bimodal bystander framework proposed by Martino and Pluchino helps reconcile these findings by emphasizing the therapeutic role of undifferentiated progenitors in neuroprotection and immune modulation, but also underscores safety concerns related to long-term engraftment. Moving forward, careful optimization of cell source, differentiation control, transplantation paradigms, and rigorous reporting of both positive and negative data will be essential to realize the clinical potential of NSC-based approaches for AD.