Senescent brain cell types in Alzheimer's disease: Pathological mechanisms and therapeutic opportunities

Abstract

Cellular senescence is a cell state triggered by programmed physiological processes or cellular stress responses. Stress-induced senescent cells often acquire pathogenic traits, including a toxic secretome and resistance to apoptosis. When pathogenic senescent cells form faster than they are cleared by the immune system, they accumulate in tissues throughout the body and contribute to age-related diseases, including neurodegeneration. This review highlights evidence of pathogenic senescent cells in the brain and their role in Alzheimer's disease (AD), the leading cause of dementia in older adults. We also discuss the progress and challenges of senotherapies, pharmacological strategies to clear senescent cells or mitigate their toxic effects, which hold promise as interventions for AD and related dementias (ADRD).

Introduction

Advancing chronological age is the greatest risk factor for developing Alzheimer's disease (AD), regardless of whether it is inherited (familial AD (FAD) or occurs sporadically (late-onset AD, LOAD) [1]. Studying the mechanisms of biological aging may lead to a greater understanding of pathogenic processes driving AD and related dementias (ADRD) [2]. Twelve hallmarks of biological aging have been identified as key mediators of healthy aging, and their perturbation contributes to chronic age-associated diseases and functional decline [3]. Several of the hallmarks of aging have been implicated in AD pathophysiology [2]. For example, dysfunctional proteostasis and impaired autophagy contribute to the formation, persistence and spread of hallmark AD pathologies, intracellular tau aggregation and extracellular amyloid-β (Aβ) deposition [4,5], while mitochondrial dysfunction has been proposed as an upstream driver of disease onset and progression [6,7]. In recent years, epigenetic alterations, genomic instability, inflammation and cellular senescence have all emerged as important mediators of AD pathogenesis [[8], [9], [10], [11], [12]]. Cellular senescence is particularly intriguing as it reflects a cellular state that encompasses and integrates the other biological aging hallmarks: genomic instability, telomere attrition, epigenetic alterations, loss of proteostasis, altered macroautophagy, mitochondrial dysfunction, aberrant nutrient sensing, stem cell exhaustion, altered intercellular communication, dysbiosis and chronic inflammation.

Cellular senescence arises through a multi-step process that is characterized by an end-stage, firm cell cycle arrest and resistance to apoptosis [13,14]. Cells that enter senescence through physiological, programmed processes serve functional roles during development, wound healing, tissue repair and the formation of blood-tissue barriers [[15], [16], [17]]. In contrast, stress-induced premature senescence (SIPS) arises through complex, dynamic stress responses that allow damaged cells to escape cell death and survive [[18], [19], [20]]. The stress responses leading to pathogenic SIPS can be triggered by a variety of endogenous or exogenous stressors, including telomere attrition, DNA damage, lipid accumulation or proteotoxic stress [21] and are accompanied by epigenetic changes, mitochondrial dysfunction, aberrant proteostasis and chronic inflammation [22]. Mitochondrial dysfunction in senescent cells often drives the release of pro-inflammatory factors collectively termed the senescence-associated secretory phenotype (SASP) [[23], [24], [25]]. The SASP includes chemokines that recruit immune cells, which play an important role in clearing senescent cells. However, if the immune system fails to remove senescent cells, components of the SASP cause chronic inflammation, become toxic to the microenvironment, and may promote neighboring cells to either enter senescence or undergo cell death. When stress-induced pathogenic senescent cells accumulate, they contribute to tissue dysfunction and diseases, including ADRD [11,12,[26], [27], [28], [29], [30]].

A common mechanism underlying SIPS is the activation of senescent cell anti-apoptotic pathways (SCAPs), which enable the survival of damaged cells by overriding pro-apoptotic signals [31]. The final senescent cell state, referred to as G_X, represents a stable cell cycle arrest that can occur at various stages of the cell cycle [32]. The variability in the type and duration of senescence-inducing stress, the SCAP pathways, the cell cycle stage preceding G_X arrest, and the cell type involved results in significant phenotypic heterogeneity among senescent cells [33,34]. Without universal biomarkers for senescence, detecting senescent cells in tissues is notoriously difficult. Some features of senescent cells include enlarged, aberrant cell morphology; dysfunctional lysosomes and mitochondria; DNA damage response; apoptosis resistance; and a hypersecretory phenotype. Recent reviews have proposed criteria for identifying senescent brain cells, all of which highlight the need for multiple, overlapping criteria and careful interpretation of the data [23,32,35].

Growing recognition of the role of senescent cells and their SASP in chronic inflammation and the pathophysiology of many age-associated diseases, including AD, has spurred efforts to develop pharmacological interventions that either clear senescent cells (senolytics) or suppress their inflammatory effects (senomorphics) [11]. Proof-of-concept studies in laboratory models, including those with AD pathology, suggest that senotherapeutics can improve brain structure and function, establishing a causal link between cellular senescence and ADRD [[36], [37], [38]]. This review provides a concise overview of current knowledge in this emerging field, highlighting opportunities for advancement, challenges in identifying senescent cells, and recent progress in senotherapeutic approaches to better understand the role of cellular senescence in brain aging and ADRD.

Cellular Senescence in the Alzheimer's Disease Brain

A role for senescent cells in AD pathophysiology has been established in recent years [[36], [37], [38], [39], [40], [41], [42], [43], [44], [45], [46]]. Studies have determined that hallmark AD pathologies, intracellular tau-containing neurofibrillary tangles (NFTs) and extracellular Aβ plaques, are mechanistically linked to cellular senescence [[36], [37], [38]]. Both post-mitotic and mitotically active brain cells have been shown to undergo senescence and contribute to neuronal dysfunction, disrupted glial homeostasis, blood-brain barrier (BBB) impairment, and chronic neuroinflammation. Additionally, emerging evidence suggests that LOAD risk alleles, such as APOE4 and TREM2-R47H, predispose cells to enter senescence [46], providing a potential mechanism by which genetic risk factors drive disease through a hallmark of aging [47]. Furthermore, SASP factors have been detected in the blood and cerebrospinal fluid (CSF) of AD patients [48]. While it remains unclear whether these factors originate from senescent cells in the central nervous system (CNS) or peripheral blood and tissues, they suggest the presence of senescent cell accumulation in patients. This section reviews current evidence linking brain cell senescence to AD pathology, Fig. 1, with an emphasis on CNS senescent cells.

Fig. 1.

Alzheimer's disease pathogenic proteins contribute to brain cell senescence. (a) Overview of the interaction between senescent brain cells with amyloid plaques and pathogenic tau. (b–e) Detailed view of each respective cell type and senescence-associated features reported in the literature: (b) neuron, (c) microglia, (d) oligodendrocyte/oligodendrocyte precursor cell, (e) astrocyte, and (f) blood-brain barrier (BBB) featuring endothelial cells, pericytes, and astrocytes, demonstrating compromised BBB integrity in AD. Mitochondrial dysfunction has been reported in all cell types and is not included for simplicity. Abbreviations: SASP: senescence-associated secretory phenotype; CDKs: cyclin-dependent kinases; CDKIs: cyclin-dependent kinase inhibitors; SA β-gal: senescence-associated beta-galactosidase; TREM2: triggering receptor expressed on myeloid cells 2. Figure created with Biorender.com.

Neurons and neurescence

Neurons are the fundamental signaling cells of the nervous system, responsible for storing, processing, and retrieving information that underlies perception, learning, memory, and behavior. They are formed from neural progenitor cells during development, differentiate and remain in the G₀ quiescent phase of the cell cycle for the lifetime of the organism. Structurally, neurons consist of a cell body (soma), dendrites that receive input, and an axon that transmits signals to other cells. These highly specialized and diverse cells exhibit variations in morphology, function, and connectivity, tailored to their roles in the nervous system.

Neuronal health relies on support from glial and vascular cells to regulate ion balance, recycle neurotransmitters, provide metabolic and immune support, remove debris, remodel synapses, and ensure nutrient and oxygen delivery. If neuronal homeostasis is disrupted either intrinsically or through dysregulated support cells, stress may cause terminally differentiated neurons to exit G₀ quiescence [49,50]. Neurons that exit G₀ may undergo apoptosis, or transiently engage steps of the apoptotic pathway to reconcile damage and resist cell death [51]. Neurons often enter G₁ to repair DNA damage, subsequently returning to G₀ upon completing the repair [49,50]. Neurons that re-enter the cell cycle, but do not undergo apoptosis or return to G₀ may become aberrantly arrested in the cell cycle, which is a characteristic of senescent cells. We recently introduced the cell state 'G_X' to describe the senescence-associated cell cycle arrest when the specific stage of the cell cycle is unclear [32].

Neurons that enter G_X senescence are referred to as neurescent [32]. The transition from G₀ quiescence to G_X neurescence involves a loss of neuronal identity, including reduced expression of canonical neuronal genes, simplified neuritic arborization, and the retraction or loss of dendritic spines [43,52], Fig. 1b. These changes are associated with global epigenomic alterations, where DNA regions linked to developmental processes become more accessible [42,52]. These epigenomic alterations may simultaneously impact multiple senescence-associated mechanisms including genomic instability, persistent DNA damage, cell cycle arrest, and resistance to apoptosis [53]. To note, epigenetic reprogramming regulates SASP expression in other senescent cell types and contributes to the epigenetic modulation of immune-related genes, facilitating immune evasion, a process reminiscent of how tumor cells evade immune recognition [54,55]. Other studies have demonstrated that neuronal epigenomic changes can disrupt gene expression critical for neuronal connectivity and communication, leading to neurodegeneration and cognitive deficits [56,57]. While direct evidence connecting epigenomic changes to neurescence and subsequent cognitive effects is lacking, indirect findings strongly support this as a compelling avenue for future research.

Neurescence can be induced by various AD risk factors, including advanced age, insulin resistance, mild traumatic brain injury, and alcohol use disorder [32,[58], [59], [60], [61]]. While a comprehensive evaluation for senescence susceptibility of all neuronal cell types has not been explored, one consistent finding across the early reports has been that excitatory neurons may be particularly vulnerable to neurescence [37,39,43,52]. This may reflect the high oxidative stress and DNA damage associated with excitatory signaling, or a study design bias, given that most brain regions and cell types assessed to date have focused on cortical regions with high proportions of excitatory neurons. Further research is needed to better clarify neuronal subtype-specific vulnerability to neurescence across aging and disease.

In vivo studies indicate that neurescence has detrimental functional consequences on neuronal circuitry, cognition, and behavior. For example, clearing senescent cells improves brain structure and cognitive function in mouse models of AD pathology [[36], [37], [38]], while inducing neurescence by reducing growth differentiation factor 11 (Gdf11) expression in postmitotic neurons leads to dendritic phenotypes that disrupt circuit function [62]. In vitro models have complemented these findings by providing mechanistic insights into causes and functional consequences of neurescent cells. For example, iNeurons, which are neurons generated through the direct conversion of patient fibroblasts, model aspects of neuronal aging and AD [41,42]. iNeurons from AD patients exhibit epigenetic alterations similar to those observed in neurons from AD brain tissue and display aberrant neuronal function. These features are also seen in neurescent cells caused by the loss of SATB1 or GDF11, which mimic neurescent dopaminergic and cortical neurons, respectively [62,63]. Transcriptomic analyses and cell culture media assays from these neurescent cells indicate upregulated canonical SASP factors, providing additional evidence of neurescence.

Neurons communicate through diverse mechanisms, including electrochemical signaling, which is tightly regulated to maintain proper brain function. Disruption of this signaling is emerging as a feature of neurescence, which may contribute to brain dysfunction not only through aberrant electrical activity but also through the release of a neurescent-associated secretory phenotype (NASP) alongside canonical SASP factors [32]. Early studies indicate that neurescent cells display aberrant electrical activity and may be primed to release pathogenic, aggregate-prone proteins [37,41,42,62,64,65], which may have particularly devastating consequences in AD and related tauopathies, diseases involving pathogenic tau [66]. In tauopathies, neurons accumulate post-translationally modified tau as NFTs, and can store, release and transfer pathogenic tau to other cell types [67]. Neuronal hyperexcitability, a feature of neurescence, has been shown to exacerbate tau propagation [62,[68], [69], [70]], while pathogenic tau itself drives senescence in microglia [71], vasculature [72], astrocytes [45] and neurons [37,39,43]. The strong link between pathogenic tau and neurescence in both mouse and human brains [37,39,43] raises the intriguing possibility that NFT-bearing neurons are key contributors to pathogenesis through neurescent mechanisms. Supporting evidence includes the upregulation of senescence-related pathways—such as interferon signaling, inflammatory responses, cell cycle progression, and inhibition of cell death—in NFT-bearing neurons [37]. These neurons also exhibit additional classical features of senescence, including increased expression of the cyclin-dependent kinase inhibitor (CDKI) p19^INK4d and abnormal nuclear morphology [39,73]. We speculate that neurescent cells may initiate and sustain the pathogenic spread of neurotoxic proteins across multiple cell types.

While these initial studies provide compelling evidence that neurescence may contribute to brain dysfunction through circuit disruption and the release of SASP and NASP factors, further mechanistic research is needed to fully elucidate its impact on brain aging and AD. Importantly, emerging evidence indicates that neurescence extends beyond aging and AD to other neurodegenerative and neuropsychiatric conditions, including Parkinson's disease, tauopathies, viral infections, spinal cord injury, depression and increased intracranial pressure. This growing body of research underscores neurescence as a critical factor in brain aging and disease, and we anticipate continued advancements in the field. For further details on neurescence, including its cell cycle regulation, methods for identification in tissue, and models for its spread, we recommend recent reviews [11,32,53].

Oligodendrocyte precursor cells (OPCs) and oligodendrocytes

Neurons are supported by many brain cell types, including oligodendrocytes, a glial cell type that myelinates neuronal axons of the CNS. Myelination begins during fetal development and continues into early adulthood, and exhibits plasticity in the adult brain where it contributes to experience, learning and neuronal activity. Myelinated neuronal axons of the CNS are primarily located in the white matter regions, such as the corpus callosum and internal capsule, with less presence in the grey matter. Oligodendrocyte precursor cells (OPCs) are a heterogeneous, multipotent population evenly distributed across the white and gray matter. While best known for differentiating into myelin-producing oligodendrocytes, OPCs also play key roles in neural circuit remodeling, angiogenesis, and the regulation of neuroinflammation, cognition and behavior [74,75]. As the most proliferative cells in the brain, OPCs account for more than 70 ​% of CNS dividing cells [76].

Cell-intrinsic regulatory mechanisms of OPC function include genetic and epigenetic factors that control their proliferation, differentiation, and survival. Interestingly, in artificial cell culture environments, OPCs proliferate indefinitely and only become senescent when exposed to genotoxic drugs [77]. In vivo, however, the behaviors of OPCs are also influenced by cell-extrinsic cues, chemical and mechanical signals from the extracellular matrix and neighboring cells. Age-related tissue stiffness is sufficient to impair the regenerative potential of OPCs [78]. Aging and pathology-associated changes, such as oxidative stress, pro-inflammatory cytokine release, and toxic protein aggregation, also disrupt their local niche. This harsh microenvironment may drive region-specific OPC dysfunction and senescence. For example, in the aging brain white matter resident OPCs are more prone to senescence than their hippocampal counterparts [79], a phenomenon that coincides with the regional prevalence of senescent and disease-associated microglia [80].

Studies in mice have demonstrated that OPC senescence accelerates myelin loss and disrupts immune homeostasis, creating a vicious feedback loop. In rodent models of multiple sclerosis, a demyelinating disease, senescent OPCs become unresponsive to pro-differentiation signals and are linked to irreversible neurodegeneration [81,82]. The role of senescent OPCs as upstream drivers of AD pathogenesis, however, appears unlikely as studies have identified an inverse correlation between multiple sclerosis and AD pathology [83]. Instead, data from mouse models suggest that AD pathology may be upstream of OPC senescence. For example, pathogenic tau-induced axonal damage accelerates OPC proliferation and differentiation [84,85], resulting in aberrant myelination patterns in tauopathy mouse models [86]. Similarly, changes in OPC proliferation patterns have been observed across various amyloidogenic mouse models [87]. In 3xTg mice, which develop both amyloid and tau pathologies, OPC density is comparable to age-matched controls [88]. However, as the mice aged and the disease progressed, OPC volumes increased and became significantly greater than those of age-matched non-transgenic control mice. The enlarged cell volume, a hallmark of cellular senescence, suggests that AD pathologies primarily induce OPC senescence rather than depleting them through apoptotic mechanisms.

A study evaluating postmortem human AD brain tissue reported that over 80 ​% of amyloid plaques are associated with nearby OPCs expressing senescence markers, such as senescence-associated β-galactosidase (SA β-gal) activity and CDKI expression (p16^INK4a and p21^Waf1/Cip1) [38], Fig. 1d. These data suggest that amyloid plaque environments may induce OPC senescence; however, evidence for Aβ driving OPC senescence remains unclear. Aggregating Aβ_1–42 can directly induce senescence in cultured OPCs, but this effect is observed in only ∼10 ​% of cells [8] suggesting that amyloid-independent processes also contribute to OPC senescence in AD brain tissue. Signals from disease-associated microglia and astrocytes, which are activated by amyloid and tau pathologies [89,90], may contribute to wide-spread OPC senescence observed in plaque regions in the brain. OPCs play a crucial role in regulating immune homeostasis by suppressing microglial activation [91,92]. This function may become impaired when OPCs enter senescence; consequently, OPC senescence may exacerbate neuroinflammation not only through SASP secretion but also by driving microglia into disease-associated states. The most conclusive evidence for a role of cellular senescence in amyloid pathogenesis are data indicating that senolytic removal of senescent cells, including senescent OPCs, ameliorates Aβ plaque accumulation and neuroinflammation in Tg(APPswe/PSEN1dE9) mice [38]; these mice are referred to as TgAPP/PS1 here forward. While additional studies are needed to elucidate the mechanisms by which plaque regions drive OPC senescence, data supporting the interplay among OPC senescence, AD pathophysiology, and brain dysfunction are compelling and represent an important area for future investigation.

Mature oligodendrocytes may also become senescent and contribute to neurodegeneration. The underlying molecular mechanisms include Jun activation domain-binding protein 1 (JAB1) and NF-κB signaling [93,94]. JAB1 regulates many pathways critical to cellular senescence, including cell cycle progression, apoptosis, and DNA damage and repair [95]. Jab1 deletion in transgenic mice leads to p21-associated senescence in both early-phase and mature oligodendrocytes, featuring SASP and DNA damage accumulation. The mice also develop progressive demyelination across the CNS [93]. Similarly, activation of NF-κB in mature oligodendrocytes induces senescence, resulting in diminished myelination capacity, inflammation through SASP secretion, and reactive microgliosis [94]. The role of senescent oligodendrocytes in ADRD pathophysiology has not yet been evaluated, representing a new area of research warranting investigation.

Overall, senescence in OPCs and oligodendrocytes may contribute to ADRD pathophysiology by disrupting myelination and amplifying neuroinflammation which would have negative consequences on brain health and cognitive function. Data from early studies indicate that OPC senescence may primarily represent SIPS triggered by hostile microenvironments, including AD pathology and interactions with neighboring senescent cells, and not replicative senescence. These pioneering studies also support the beneficial effects of senotherapeutics in halting amyloid deposition and ongoing demyelination while actively promoting remyelination by restoring functional oligodendrocyte pools. This strategy holds potential for application across a range of CNS disorders where myelination deficits are key pathological features, which is discussed in later sections.

Microglia

Microglia, another type of glial cell in the brain, are derived from yolk sac progenitors that migrate into the brain parenchyma during development [96]. As the resident macrophages in the brain, microglia promote tissue homeostasis by phagocytosing senescent cells, as well as other dead and dysfunctional cells, synapses of living neurons, toxic protein aggregates, and debris. Unlike other cells of hematopoietic origin, mature microglia persist and turn over at a slow rate under homeostatic conditions. In mice, the median lifetime of neocortical resident microglia is approximately 15 months [97]. In humans, some microglia persist for over two decades, with a median annual renewal rate of 28 ​% [98]. As one of the few brain cell types capable of proliferation, microglia are prone to senescence driven by replicative stress [99]. Their phagocytic activity increases their vulnerability to the intracellular accumulation of toxic waste, which can contribute to SIPS [71,100]. Their long lifespan also renders them susceptible to other chronic, senescence-promoting factors, including aging, commensal bacterial composition [80], toxic protein aggregates [99], and oxidative stress [101], Fig. 1c.

Microglia that undergo replicative senescence appear to be linked to amyloid pathology. In the Tg(APP/PS1) mouse model of amyloidosis, Aβ production triggers prolonged and excessive microglia proliferation, leading to further amyloid accumulation and emergence of disease-associated microglia (DAM), a subpopulation with distinctive transcriptional and functional signatures, recently identified in both human and mouse models of AD [8,9]. These DAM microglia also exhibit other characteristics associated with replicative senescence, including p16^INK4a upregulation, SASP factor secretion and telomere shortening [99]. However, other studies have shown that DAM and senescent microglia are not identical. In the 5xFAD mouse model of amyloidosis and double mutant K257T/P301S mouse model of tauopathy (DMhTAU), senescent microglia displayed a protein signature distinctive from DAM, including canonical senescence-related proteins, homeostatic microglial markers, and elevated levels of apolipoprotein E (ApoE) and triggering receptor expressed on myeloid cells 2 (TREM2) [102]. Transcriptional profiling of the P301S (PS19) tauopathy mice also revealed that only a subset of DAM display senescent features, and these two subgroups overlap primarily due to their shared inflammatory phenotype [103]. These findings demonstrate the heterogeneity of microglial responses to AD pathologies. Senescent microglia may reflect a later stage in the progression of DAM along a shared trajectory. Alternatively, they could represent distinct, parallel responses to stressors. Further characterizations are needed to clarify the relationship between DAM and senescent microglia.

Microglial senescence can also be driven by other stressors. In P301S (PS19) mice, live neurons harboring tau aggregates have been shown to expose phosphatidylserine on their plasma membrane, inviting microglial phagocytosis [100]. Intracellular pathogenic tau may render microglia vulnerable to stress-response pathways that induce senescence. After engulfing tau-containing neurons, the microglia became hypophagocytic, secreted insoluble tau seeds, and exhibited a senescence-like phenotype, including SA β-gal activity and secretion of SASP factors [71]. Additionally, extracellular tau stimulation induced mitochondrial DNA (mitoDNA) leakage in cultured microglia, leading to the activation of cGAS-STING pathway, an anti-viral response activated by the presence of cytoplasmic double-stranded DNA [46,104]. The cGAS-STING pathway has been shown to initiate cellular senescence in many cell types [[105], [106], [107]]. In P301S (PS19) mice, APOE4 and TREM2 LOAD risk alleles exacerbated tauopathy-induced mitoDNA-mediated cGAS-STING activation, leading to microglial senescence [46]. Taken together, these studies indicate that microglial senescence is closely linked to AD pathogenic proteins.

Overall, current evidence suggests that Aβ pathology may drive microglia senescence primarily through replication exhaustion, whereas tau may trigger other cellular responses leading to SIPS. Microglial senescence elevates chronic neuroinflammation and may promote secondary senescence through SASP. Additionally, the diminished capacity of senescent microglia for phagocytosis reduces the effectiveness in clearing dead cells, dysfunctional synapses, and myelin debris, leading to dysregulation of tissue homeostasis. It is worth noting that DAM and senescent microglia share many proteomic and transcriptomic signatures, making it difficult to identify microglial senescence using a single, definitive marker. Replicative and stress-induced microglial senescence may also differ in their transcriptional and functional profiles. Future studies are required to characterize microglial senescence at finer resolution, investigating how AD risk factors contribute to its onset, and delineate how it differs from other closely related microglial states.

Infiltrating peripheral immune cells

Microglia are the resident macrophages of the brain; however, recent evidence indicates that other immune cells also contribute to brain health and dysfunction. The concept of immune privilege in the brain has evolved significantly in recent years. The brain was once thought to be an isolated, immune-silent organ. However, borders of the CNS serve as central sites of neuro-immune interactions where the brain actively participates in immune processes [108]. The senescence of peripheral immune cells has emerged as particularly devastating to organismal health. They infiltrate the parenchyma, driving brain cells into senescence under both physiological and pathological conditions [109,110]. Additionally, circulating myeloid cells harboring senescence-inducing mutations have been shown to induce BBB breakdown, infiltrate the brain, and transition into senescent macrophages that contribute to neurodegeneration [111]. In contrast, clonal hematopoiesis of indeterminate potential (CHIP), which refers to acquired mutations in hematopoietic stem cells in the absence of blood cancer, has been associated with reduced risk of AD [112]. Many CHIP mutations are linked to elevated inflammation [113], and CHIP-carrying HSCs infiltrate the brain and adopt a microglial-like phenotype [112]. While future research is needed to fully understand the inverse correlation between CHIP and AD, the early findings add to the growing appreciation for both negative and positive interactions between infiltrating peripheral immune cells and brain health.

Brain cells harboring AD pathology also modulate the immune response in the parenchyma. A correlation between infiltrated, mature T cell and phospho-tau load has been reported in AD patients [114]. In tauopathy mice that express APOE4, pathogenic tau pushed homeostatic microglia toward a disease-associated state, accompanied by an increase in chemokine and cytokine production, resulting in the recruitment of cytotoxic T cells into the brain parenchyma [115]. Their enrichment has been highly correlated with brain atrophy and microglia density, indicating that T cells also contribute to neurodegeneration [115]. In an AD mouse model of Aβ plaque deposition, neutrophils trafficked to the brain and homed to Aβ plaques, secreted IL-17 and released neutrophil extracellular traps [116]. Depleting neutrophils or blocking their trafficking into the brain improved memory of the cognitively impaired AD mice. A similar subset of neutrophils, defined by increased interleukin (IL)-17 and IL-1 co-expression gene modules, were found to interact with microglia and be associated with cognitive impairment [117]. IL-17⁺ neutrophils from APOE4 females also upregulated immunosuppressive cytokines, IL-10 and TGFβ, and immune checkpoints, LAG3 and PD-1, which are associated with accelerated immune aging. Neutrophil-specific deletion of APOE4 in AD mice reduced the immunosuppressive phenotype, restored the microglial response to neurodegeneration, and reduced plaque pathology [117]. These studies provide evidence that APOE4 may be mechanistically linked to AD pathogenesis through its effects on immune cells, consistent with its senescence-mediating effects on microglia in LOAD [46].

Aging of the immune system, marked by senescence and chronic inflammation, negatively impacts brain health, while immune rejuvenation has demonstrated beneficial effects [118]. Although the mechanisms and specific cell types mediating these effects remain under investigation, emerging evidence highlights additional immune players in the CNS, including dendritic and natural killer cells, which influence inflammation, senescence, neurogenesis, and cognitive function [[119], [120], [121], [122], [123], [124]]. Collectively, these studies emphasize that crosstalk between the brain and the peripheral immune system modulates brain health and function. Targeting peripheral senescent immune cells and alleviating chronic inflammation represent promising strategies to improve brain health and combat neurodegeneration, perhaps without the need for therapies to penetrate the brain. As this field continues to expand, senotherapeutics may offer a novel and effective approach, as discussed in the therapeutic section.

Cells of the blood-brain barrier (BBB)

The BBB is a highly selective diffusion barrier composed of multiple interconnected cell types. This structure regulates cellular and molecular communication between the brain and periphery, including interactions between brain cells and peripheral immune cells. Endothelial cells line the brain blood vessels and form tight junctions that restrict the movement of substances from the bloodstream into the brain. Pericytes, which wrap around these endothelial cells, regulate blood flow and maintain BBB stability. Astrocytes support the barrier by extending their end-feet around the blood vessels, providing biochemical support and reinforcing barrier integrity, Fig. 1e. Together, the BBB protects the brain from harmful substances in the blood while allowing the regulated passage of essential nutrients and signaling molecules. Disruption or dysfunction of BBB cells, such as cellular senescence, contributes to various diseases of the CNS, including ADRD.

Senescence of various BBB cell types has been observed in both aging and AD, Fig. 1f. Gene network analysis revealed that, among all brain cell types analyzed, vascular smooth muscle cells preferentially expressed senescence-related genes [125]. However, since this study used healthy human tissue, it remains unclear whether these senescent barrier cells were dysfunctional. Interestingly, some evidence suggests that tissue-barrier cells may exhibit physiological, rather than pathological, senescence signatures. For example, transcriptomic studies of postmortem human brains found a high prevalence of senescence signatures in BBB-associated cells from fetal brains [39], suggesting that these cells may represent a physiological rather than pathological cell type. Similarly, barrier cells in the bladder have been shown to express a physiological senescence phenotype, underscoring that not all senescent cells are harmful [126].

Cellular senescence of other cell types comprising the BBB may contribute to compromised structural and functional integrity of the barrier. For example, cerebral microvessels isolated from human AD prefrontal cortex with high pathogenic tau deposition displayed a strong senescence transcription profile [127]. These include the upregulation of factors that attract circulating leukocytes, promoting their adhesion and potentially allowing peripheral immune cell infiltration [127]. The microvessel isolates were confirmed to be enriched for endothelial cells and smooth muscle cells; however, given that microvessel composition also includes pericytes, smooth muscle cells, and fibroblasts [128,129], it would be interesting for future studies to disentangle which specific cell type(s) within the microvessels contribute to the senescence signature. Notably, pathogenic tau has been associated with brain endothelial cell senescence in P301S (PS19) tauopathy mice [72]. Senescent endothelial cells were attributed to impaired vasodilation and restricted cerebral blood flow in the mice. In Tg(APP/PS1) amyloidogenic mice, senescent endothelial cells were found at sites of vascular leakage, characterized by significantly downregulated adhesion molecule pathways. This change in gene expression could lead to further BBB breakdown [130]. Astrocytes also adopt a senescence-like phenotype in AD and other tauopathies, including frontotemporal dementia. Mechanistically, oligomeric tau has been shown to induce astrocyte senescence, characterized by DNA damage, mislocalization and release of high mobility group box 1 (HMGB1), and increased inflammation [45,131]. Similarly, amyloid pathologies cause the upregulation of the G-protein-coupled adenosine receptor, A_2AR in astrocytes [132], which has been demonstrated to modulate astrocytic senescence in the hippocampus [133]. Given that astrocytes serve additional functions beyond maintaining the BBB, further work is needed to understand the mechanisms by which astrocyte senescence may contribute to AD pathogenesis.

Collectively, these initial studies on cellular senescence in BBB cells suggest that it may play a significant role in brain dysfunction associated with aging and ADRD. Given that pathogenic tau is associated with senescence across various BBB cell types, a critical question emerges: how does this neuronally expressed protein drive such widespread senescence across different cell types? We propose that neuronal tau accumulation may serve as a central trigger of neurescence, which spreads senescence to other cell types, thereby contributing to brain dysfunction in aging and disease [32], [134]. Investigating these mechanisms further could uncover novel therapeutic strategies targeting tau-induced senescence, with the potential to preserve BBB integrity in neurodegenerative conditions.

Senotherapeutics and Their Impacts on Brain Structure and Function

Therapeutic strategies to mitigate the detrimental effects of senescent cells, referred to as senotherapeutics, are rapidly advancing, including for conditions of the CNS [11]. These strategies encompass senolytics (agents that selectively clear senescent cells) [31], senomorphics (agents that reduce the harmful effects of the SASP), and cellular reprogramming to alter the senescent cell state [136,137]. Among these, senolytics are the most advanced in terms of clinical translation and are the primary focus of this section, Fig. 2. For a broader and more in-depth discussion of senotherapeutics in the brain, we refer readers to a recent comprehensive review [11].

Fig. 2.

Mechanisms and applications of senotherapeutics targeting senescence in the brain. (a) Simplified schematic representation of a cell maintaining its senescent state through engaging senescent cell anti-apoptotic pathways (SCAPs). Senolytics such as dasatinib, navitoclax, quercetin, and fisetin inhibit SCAP pathway signaling, thereby promoting senolysis. Senomorphics inhibits the senescence-associated secretory phenotype (SASP). (b) Overview of senolytics currently in clinical trials and senescent cell types they may target as evidenced from laboratory studies. Abbreviations: D+Q: dasatinib plus quercetin; RTK: receptor tyrosine kinase; SASP: senescence-associated secretory phenotype; HSPs: heat shock proteins; OPC: oligodendrocyte precursor cell. Figure created withBiorender.com.

Senescent cells evade cell death by activating SCAPs, which include the upregulation of anti-apoptotic proteins such as BCL-2 and BCL-XL and their upstream regulators, such as HSP-90, PI3K, and p53 [31]. By targeting SCAP pathways, senolytics shift the balance toward apoptosis, enabling the selective clearance of senescent cells, Fig. 2a. While the direct effects of senolytics on brain cells are under active investigation, their ability to target senescent cells in peripheral tissues, including the peripheral immune system, may also confer benefits to brain health. As discussed above, peripheral immune cell senescence has been shown to exacerbate neuroinflammation and cognitive decline through systemic inflammation and immune dysregulation, suggesting that senolytics could have dual benefits by improving both systemic and brain health. This section explores the impact of senolytics on various brain cell types, Fig. 2b, with a focus on their potential for clinical application in mild cognitive impairment (MCI) and AD [138].

Dasatinib with quercetin, D + Q

Dasatinib is a potent, broad-spectrum tyrosine kinase inhibitor primarily targeting the BCR-ABL and SRC family kinases [[139], [140], [141]]. It was originally developed to inhibit the abnormal BCR-ABL fusion protein that drives chronic myeloid leukemia and Philadelphia-chromosome-positive acute lymphoblastic leukemia. It also blocks several other kinases involved in cell signaling, growth, and survival, making it an appealing candidate for targeting senescent cells [31,142,143]. In a cell culture screen of compounds hypothesized to induce senolysis, the combination of dasatinib with quercetin, a non-specific PI3K inhibitor with anti-inflammatory effects (referred to as D + Q), acted synergistically to induce apoptosis in various types of senescent cells and proved effective in vivo [31].

The senolytic activity of D ​+ ​Q has been observed in various laboratory models of AD. For example, in iPSC-derived neurons from AD patients, D ​+ ​Q effectively induced cell death and selectively removed p16^INK4a-expressing senescent cells [41]. In rTg(tau_P301L)4510 tau transgenic mouse model, D ​+ ​Q reduced the number of NFT-bearing neurons and concomitant SASP-related gene expression, restored neuron number and aberrant cerebral blood flow [37]. In amyloid mouse models, D ​+ ​Q has been shown to clear senescent OPCs surrounding Aβ aggregates and autophagy-deficient senescent microglia, alleviate plaque-related pathologies, and alter cognition with sex-specific benefits [38,144,145].

Following the preclinical evidence that senolytics improved brain structure and function in laboratory models of AD, and confirmed reports of BBB penetrance of D + Q [[146], [147], [148], [149]], a Phase I trial enrolled older adults with early AD with the primary goal of evaluating safety and BBB penetrance [150,151]. The treatment was well-tolerated, with no early discontinuations. Dasatinib was detected in the CNS of 4/5 subjects with levels positively correlating with CSF neurofilament light, a marker of neurodegeneration [151,152]. While quercetin has shown benefits for brain health and function, and demonstrated brain penetrance in laboratory models [146,147], the trial was unable to detect quercetin or its metabolites in the CSF after six rounds of intermittent D ​+ ​Q treatment [151]. While the reasons are unclear, potential explanations may include that quercetin did not penetrate the brain due to its low bioavailability and known limited BBB penetration. Alternatively, quercetin may have penetrated the brain but went undetected due to limitations in instrument sensitivity, or because its presence was not captured, given its short half-life. Nevertheless, secondary and exploratory outcome data from the trial suggested that peripheral senescence and AD pathogenic factors may be ameliorated by the treatment, which supported Phase II testing [30,151]. Additional independent clinical trials are underway to explore the potential of D ​+ ​Q for MCI, AD, and treatment-resistant depression [12]. We anticipate that within the next few years, the neuroscience field will have a better understanding of the therapeutic potential for the use of D ​+ ​Q in multiple conditions of brain aging and dysfunction.

Fisetin

Fisetin, a naturally occurring flavonoid structurally similar to quercetin, is currently in clinical trials for various geriatric conditions, including multimorbidity (NCT06431932) and vascular dysfunction (NCT06133634). It has diverse pharmacological effects, including antioxidant, anti-inflammatory, and senolytic properties. Fisetin modulates several senescence-maintaining pathways, including but not limited to mTOR, PI3K/Akt, and BCL-2 [153]. Fisetin treatment has been shown to reduce senescence across various peripheral tissues in progeria mice and extend their lifespan [154]. Regarding its utility in modulating senescent brain cells, fisetin has been shown to eliminate senescent dopaminergic neurons derived from human iPSCs in vitro [63]. Additionally, due to its broad biological effects, fisetin improves neuronal function independently of its senotherapeutic activity. In vitro studies have demonstrated that fisetin reduces markers of oxidative stress and tau aggregation in neurons [155,156]. In the Tg(APP/PS1) amyloid mouse model, fisetin suppressed p25/CDK5 signaling, leading to reduced astrocytic reactivity and neuroinflammation [157]. Single-cell transcriptomic analysis showed that Cdk5r1, the gene encoding p35 (the precursor of p25), is primarily expressed in neurons [158]. Current findings suggest that fisetin primarily targets neurons, addressing their dysfunction and, in turn, alleviating secondary glial toxicity. In a separate study using Tg(APP/PS1) mice, fisetin demonstrated strong anti-inflammatory effects in male mice suggesting sex-dependent responses to senolytics [145]. These studies demonstrate that, in addition to its senolytic effects, fisetin likely exhibits anti-inflammatory senomorphic properties and underscores the need to better understand sex differences in the senotherapeutic response.

The pharmacokinetics of fisetin across the BBB are not yet fully understood. A study using in vitro BBB models suggests that fisetin might have high potential to penetrate the barrier [159]. Following intraperitoneal administration of liposomal fisetin formulations to rats subjected to ischemic shock, brain levels reached 8.23 ​ng/g [160]. However, it is yet to be determined whether this concentration is sufficient to effectively induce senolysis or modulate brain cell functions. Given the diverse benefits of fisetin on organismal health and its favorable safety profile, further research is needed to fully understand its senotherapeutic potential. This should include a deeper mechanistic understanding of senolytic versus senomorphic properties, BBB penetrance, sex-specific responses, and the pathways through which fisetin modulates senescent neuronal and glial populations and inflammation in vivo.

BCL-2 inhibitors

Navitoclax (ABT-263) inhibits BCL-2 family proteins, shifting the balance toward pro-apoptotic signaling. While navitoclax exhibits weak BBB penetrance, with a 2–3% brain/plasma ratio in rats [161], its potential effects on brain senescent cells cannot be entirely ruled out. Age- and pathology-related increases in BBB permeability may allow greater access to the brain [162]. Moreover, navitoclax may exert benefits by clearing senescent cells in peripheral tissue, such as reducing systemic inflammation and improving immune function, without requiring direct brain penetration, potentially benefiting brain health indirectly.

In vivo studies have demonstrated that navitoclax clears a variety of senescent brain cells. In middle-aged wild-type mice, navitoclax removed senescent neural progenitor cells in the hippocampal dentate gyrus, thereby restoring adult neurogenesis potential [163]. In the P301S (PS19) mouse model of tauopathy, navitoclax reduced p16^INK4a-expressing senescent microglia and astrocytes, thereby attenuating NFT formation [36]. The authors did not examine neurescence or the potential interaction with navitoclax on neurons with or without neuropathology. However, in a separate study, ABT-737, another BCL-2 family inhibitor, exhibited toxic effects to iPSC-derived dopaminergic neurons [63]. This may have resulted from the inhibition of BCL-X_L, which promotes neuronal survival during development [164]. Similarly, BCL-X_L signaling is important to maintain the lifespan of circulating platelets. Thrombocytopenia, resulting from on-target inhibition of BCL-X_L, has been reported as a side effect of navitoclax [165,166]. These observations highlight that anti-apoptotic BCL-2 family proteins are not unique to senescence but are essential for the survival of many cell types. These on-target, off-cell effects need to be carefully monitored as they could result in harmful side effects. A more comprehensive assessment of the BCL-2 inhibitors is necessary to determine their utility and safety as viable therapeutics for brain cell senescence.

Successes and Obstacles of Senotherapy

Senotherapeutics show early promise due to their benefits across multiple laboratory models and favorable safety profiles in early human trials [151,167,168]. Unlike traditional approaches that primarily manage symptoms or target AD pathology, senotherapeutics offer a novel strategy that potentially alters the biology underlying ADRD. D ​+ ​Q combination therapy has demonstrated a favorable safety profile in older adults with symptomatic AD [151], prompting Phase II clinical trials to assess senolytic activity and monitor senescent cell re-emergence over a 12-month period in a randomized, double-blind, placebo-controlled trial [30]. Several important questions remain unanswered in this emerging field. For example, the lack of robust senescence biomarkers presents obstacles for participant selection and target engagement measurement in clinical trials. The development of biomarkers for cellular senescence is challenging due to its heterogeneity, as different cell types require distinct and often multiple phenotypic measures [32]. In preclinical studies, reliable identification of senescence involves a combination of histological staining, transcriptional analysis, and organelle functional assays—methods that are impractical in clinical settings. Additionally, senescent cells exert diverse effects on the surrounding microenvironment, many of which overlap with other inflammatory processes, complicating the development of specific SASP panels for evaluating target engagement and treatment efficacy.

In the follow-up analysis of the Phase I D ​+ ​Q combination therapy, a broad range of biomarkers from biofluids (plasma, CSF, and urine) were assessed via immune assays, mass spectrometry, lipidomics, and transcriptomic profiling. Among all detection platforms, the most significant changes were observed in lipidomics of blood and transcriptomic analysis of peripheral blood mononuclear cells (PBMCs) [152]. After treatment, plasma levels of circulating lipid species, including phosphatidylcholine, lysophosphatidylethanolamine, and acylcarnitine, were reduced. Similar changes were not observed in the CSF, which could reflect lower drug exposure, differential effects of senolytics on periphery versus central tissues, or be a result of the short treatment duration and small sample size. Future studies incorporating a placebo control group and extended treatment duration will help establish whether lipidomics is a relevant biomarker of brain-specific senolytic effects. Nevertheless, these results highlight the sensitivity of lipidomics in assessing D ​+ ​Q-associated effects in the blood. Transcriptomic analysis of PBMCs showed reduced levels of chronic stress markers, including FOSB, PTGS2, IL8, FOS, IL1β, JUNB, and JUN, all associated with senescence and SASP secretion, which overlap with the conserved transcriptional response to adversity (CTRA) [169]. Given that stress and peripheral inflammation are both potential risk factors for AD [109,170], PBMC transcription analysis, such as the CTRA, holds promise for providing an accurate reflection of senolytic efficacy.

Limited knowledge exists regarding the time course of senescent cell re-emergence after clearance in human subjects. Therefore, the frequency of dosing will need to be experimentally determined, requiring reliable outcome measures that reflect senescent cell burden. The Phase II SToMP-AD clinical trial will provide insights into this question by collecting blood samples at 3, 6, and 9 months following the last dose of D ​+ ​Q [30]. Senolytics target cell pro-survival pathways, placing them at risk for potential toxic side effects. An intermittent dosing protocol has been employed universally in all current studies; this strategy is hypothesized to reduce the potential for side effects. Notably, a favorable safety profile has been reported for all studies using D ​+ ​Q to date. However, additional clinical trials are needed to establish safety of this strategy over a longer trial duration as the long-term effects of senolytics remain unknown.

Sex-specific effects of senolytics also remain poorly understood. In a long-term study where D ​+ ​Q was administered to young adult mice, senolytics improved the SASP profile in male mice, but exacerbated inflammation in female mice [145]. D ​+ ​Q treatment also worsened cognitive function, as assessed by novel object recognition, in female mice. Interestingly, fisetin in this study demonstrated stronger anti-inflammatory effects than D ​+ ​Q, but only in male mice. These findings highlight the need to examine the interaction of specific senolytic agents with biological sex. Recent evidence also suggests that AD genetic risk factors, like APOE4 and TREM2-R47H, predispose microglia to senescence, especially in female mice [46,47]. These initial results offer valuable initial insights into the influence of genetic risk on senescence onset, a topic that remains largely unexplored in senescence biology.

Concluding Remarks and Future Directions

Cellular senescence in the brain has recently emerged as an important area of research, offering new insights into the underlying biology of brain aging and disease. This field also opens potential avenues for developing novel treatment strategies for ADRD and other neurodegenerative conditions. Multiple brain cell types can enter senescence through physiological or stress-induced pathways, emphasizing the need to better understand the upstream drivers, downstream consequences, and cell type-specific features of these processes. Senotherapeutics have shown promise for delaying, preventing, or treating neurodegenerative diseases while supporting brain health and cognitive function throughout aging in laboratory models. These therapies may act directly by targeting senescent cells in the brain or indirectly by clearing senescent immune cells, peripheral cells or cells within the BBB. As senotherapeutics advance through clinical research, several priorities must be addressed. Developing reliable biomarkers for participant selection, target engagement in both the CNS and periphery, and evaluating treatment efficacy is essential. Additionally, rigorous clinical trials are needed to establish the long-term safety, efficacy, and potential side effects of these therapies. Early discoveries underscore the importance of understanding cellular senescence in the brain. Continued investigation is necessary to fully elucidate its implications for brain health and disease, paving the way for transformative advancements in aging-related research and therapeutic development.

Author Contribution Statement

HRH and XS contributed equally to the first draft of the manuscript. HRH constructed the figures. MEO provided critical edits to the manuscript. All authors agreed upon the final version of the manuscript.

Declaration of competing interest

MEO has a patent pending, ‘Detecting and Treating Conditions Associated with Neuronal Senescence. The other authors declare no competing interests in relation to this work.