Inflammation and DNA methylation in Alzheimer’s disease: mechanisms of epigenetic remodelling by immune cell oxidants in the ageing brain

ABSTRACT

Alzheimer’s disease is a neurodegenerative disease involving memory impairment, confusion, and behavioural changes. The disease is characterised by the accumulation of amyloid beta plaques and neurofibrillary tangles in the brain, which disrupt normal neuronal function. There is no known cure for Alzheimer’s disease and due to increasing life expectancy, occurrence is projected to rise over the coming decades. The causes of Alzheimer’s disease are multifactorial with inflammation, oxidative stress, genetic and epigenetic variation, and cerebrovascular abnormalities among the strongest contributors. We review the current literature surrounding inflammation and epigenetics in Alzheimer’s disease, with a focus on how oxidants from infiltrating immune cells have the potential to alter DNA methylation profiles in the ageing brain.

Introduction

Alzheimer’s disease (AD) is the most common form of dementia, comprising 60–70% of global cases. Since there are no existing curative treatments, occurrence is expected to increase dramatically over the next thirty years, particularly as humans are living longer [1,2]. The underlying mechanisms of AD pathogenesis are complex and poorly understood, although a range of biological and environmental contributors have been identified. Molecular characteristics of AD are the presence of amyloid beta (Aβ) and neurofibrillary tangles (NFT), which are considered the primary pathological hallmarks [3]. Sequential cleaving of amyloid precursor protein (APP) by the β-secretase (β-site amyloid precursor protein cleaving enzyme 1, BACE1) and ỿ-secretase enzymes leads to the accumulation of Aβ peptides, which aggregate and form plaques. Hyperphosphorylation of microtubule associated protein tau (MAPT) leads to the accumulation of NFT in the neocortex, which is positively correlated with the severity of dementia [4,5]. These structures damage neurons and neural networks involved in cognitive processes such as memory and learning, contributing to the cognitive impairment observed in AD [6].

AD can be classified as either ‘early-onset’ in individuals younger than 65 years of age or ‘late-onset’. In heritable/familial AD, early onset disease coincides with mutations in APP genes, presenilin-1 (PSEN1) or presenilin-2 (PSEN2) [7] that code for proteins involved in the generation and processing of Aβ. However, the majority of individuals with severe AD present with the sporadic form of AD, or ‘late-onset’, which appears to be due to a combination of multiple genetic, molecular and environmental factors, resulting in a more complex pathogenesis and presentation than early onset disease [8–11]. The aetiology of late onset AD has been linked to polymorphisms in the apolipoprotein E (APOE) gene [12]. APOE has three main allelic variants, APOE ϵ2, APOE ϵ3, and APOE ϵ4, and the corresponding protein isoforms differentially influence plasma lipid metabolism, signalling, maintenance and repair of the central nervous system and regulation of the blood–brain barrier (BBB). In particular, the APOE ϵ4 subtype intensifies cholesterol dysregulation, stimulates inflammation, promotes metabolic dyshomeostasis [13–16], and enhances blood–brain barrier breakdown [17–24]. Recent research indicates that APOE4 homozygosity may represent a genetic form of AD, where the AD pathology manifests as early as age 55 with symptoms appearing by age 65 [25]. The prevalence and progression of Alzheimer's disease (AD) differs between men and women, with women generally exhibiting a higher incidence and more rapid progression of the disease. These differences may be partly attributable to sex-specific factors that influence both epigenetic and inflammatory processes and have been extensively reviewed elsewhere [13–16].

Epigenetic modifications are increasingly recognised as key contributors in the development and progression of AD, where alterations in DNA methylation, histone modifications, and non-coding RNA expression have been implicated in the dysregulation of gene networks [10]. Given their potential to provide insights into the complex interplay between genetic and environmental factors, elucidating epigenetic contributions in AD remains an important challenge for gaining a more complete understanding of pathogenesis.

Advanced age remains the most important individual risk factor associated with late onset AD [26]. Chronic low-level inflammation, termed ‘inflamm-ageing’, is a hallmark of ageing and is strongly associated with age-related disease [27]. Oxidants are produced by immune cells in response to infection [28]. When oxidant levels exceed antioxidant capacity, this leads to oxidative stress and cellular damage. The mammalian brain is particularly sensitive to oxidative stress due to the high oxygen consumption of brain cells, coupled with proportionately insufficient antioxidant defence systems [29]. While many studies have focussed on the cytotoxic qualities of immune-derived oxidant species, few have explored the more subtle effects of oxidants as chemical messengers. Emerging evidence supports the observation that oxidative stress can cause alterations to epigenetic patterns, such as DNA methylation and may be relevant as a mechanism for the development and/or progression of AD [28,30–32]. This review focuses on potential mechanisms of epigenetic changes in AD that link inflammation, ageing, oxidative stress, and methylomic alterations as pathological features.

Age-related systemic inflammation and immune dysfunction in Alzheimer's disease pathogenesis

Inflammation is the body’s response to danger signals and can be triggered by pathogenic invasion, tissue damage or the presence of toxins. Chronic inflammation occurs when the immune system fails to resolve acute inflammation and undergoes a continuous inflammatory response in the absence of any obvious threat. Chronic, sterile, low-grade inflammation gradually increases with advancing age and results in the prolonged production of oxidants, cytokines and microbicidal factors that further exacerbate the inflammatory response and cause collateral tissue damage [33]. A common feature in aged individuals is a preferential shift in hematopoietic stem cell (HSC) differentiation that favours an increase in the pool of circulating blood cells of the myeloid lineage (myeloid bias) [34]. An increase in circulating myeloid cells, including granulocytes, monocytes, macrophages, and dendritic cells, is considered a major driver of the immune dysfunction, decreased adaptive immunity and increased incidence of myeloproliferative disease observed in the ageing population [35]. In addition, impairment of optimal immune system function, termed immunosenescence, is observed with advancing age [36–38]. This phenomenon leads to the accumulation of defective immune cells that lack adequate self-repair and maintenance mechanisms and hence die more rapidly. Impaired phagocytosis by blood monocytes is observed with increased age and the resultant build-up of cellular debris is thought to perpetuate the characteristic chronic low-grade inflammation [37,39–42]. Considerable evidence implicates an imbalance of pro-inflammatory cytokines and proteins as major contributors to immune dysregulation in the aged. For example, IL-6, IL-1β, IL-1α, TNFα, MPO and C-reactive protein are all markers of inflamm-ageing and age-related disease [43–48]. Aged HSC exhibit inflammatory cell surface markers such as P-selectin even in the absence of diagnosed inflammatory disease, suggesting that aged HSC are exposed to an inherent inflammatory environment [49,50]. Oxidative stress caused by the cells that make up the bone marrow niche, such as endothelial cells, has been shown to regulate aged HSC fate, haematopoiesis and myeloid lineage bias [51]. However, oxidative stress in HSC can also arise from within the cell through metabolic processes including mitochondrial metabolism.

Mitochondrial contributions to immune dysfunction in AD

Mitochondria are essential organelles that are traditionally known for their central roles in bioenergetics and the regulation of cell death pathways, but they but they also perform less-recognised functions with significant implications for immune regulation. Beyond bioenergetics, mitochondria are involved in cytokine production, the release of mitochondrial DNA (mtDNA), and the differentiation of immune cells. These processes are crucial for the regulation of immune responses, and mitochondrial dysfunction can severely impact immune homeostasis.

Mitochondrial dysfunction is a complex phenomenon characterised by decreased mitochondrial energy production and increased oxidative stress. Mitochondrial dysfunction increases with ageing and is particularly detrimental in the brain, where the maintenance of a pool of optimally distributed mitochondria is inherently challenging due to both the high functional energy demands and cellular morphology of the neuron [52].

Defective mitochondria are a significant factor in late onset AD pathology, with evidence implicating disrupted mitochondrial dynamics (including fission and fusion processes) [53], impaired electron transport chain complex activity, and increased oxidant production as key contributing features [54–56]. The imbalance between these processes leads to the accumulation of damaged mitochondria, further exacerbating oxidative stress and energy deficits. Swerdlow et al. suggested that mitochondrial dysfunction is a primary event in the development of AD and proposed the ‘mitochondrial cascade hypothesis,’ highlighting the role of mitochondrial abnormalities in initiating and propagating AD pathogenesis [57]. Since then, evidence linking mitochondrial dysfunction to AD has increased significantly and has been extensively reviewed elsewhere [58–62].

Mitochondrial dysfunction may play a pivotal role in ageing immune systems. In order to maintain a quiescent state, HSC must keep metabolic activities low and rely heavily on glycolytic pathways for energy production [63]. Although HSC contain a large number of mitochondria, the majority remain essentially inactive [64]. The ageing process itself results in a characteristic increase of oxidant levels in HSC, and much of this is derived from mitochondrial sources [65,66]. One study has shown that mitochondrial membrane potential (MMP) in HSC was a better indicator of gene transcriptional states related to ageing than chronological age [67]. HSC with low MMP had low transcriptional output and upregulation of genes related to inflammatory signalling, apoptosis and DNA repair pathways, while high MMP HSC were enriched for genes related to transcriptional rate, RNA turnover and T-cell programs [67]. Another factor in the maintenance of quiescence in HSC is the ability of the cell to consistently cull mitochondria in order to keep endogenous oxidant levels low [63]. Ageing HSC can exhibit a build-up of mitochondria due to impaired mitophagy [68]. One study has shown that impairment of mitophagy in murine HSC increases sublethal levels of mitochondrial oxidants and results in loss of quiescence and increased pro-myeloid differentiation through an epigenetic mechanism [69]. Pertinently, oxidised mitochondrial DNA has been proposed as a biomarker for hematopoietic stem cell malignancies such as myelodysplastic syndromes, which are diseases often associated with advancing age [70,71].

Furthermore, one consequence of environmental stress is the activation of inflammasomes, protein complexes that activate caspase-1-regulated pro-inflammatory pathways [72]. The NLRP3 inflammasome is directly activated by oxidised DNA released by mitochondria [73]. Activation of the NLRP3 inflammasome results in the secretion of interleukin-1β as well as the extracellular release of mitochondrial DNA, which are known pro-inflammatory mediators and are implicated in AD pathology [74–77]. A recent in-depth proteomic analysis of plasma proteins in a large middle-aged human cohort has highlighted that growth differentiation factor 15 (GDF-15) may be a strong predictive biomarker of dementia and cognitive decline [78]. Whilst GDF-15 was not specifically associated with AD in this study, it is a stress-response mitokine that is also a marker of mitochondrial dysfunction [79]. GDF-15 is involved in numerous immunomodulatory activities [80–82] and should not be discounted as a central player in directing inflammatory responses and informing levels of mitochondrial dysfunction/damage in AD [83].

Microglia: neuroinflammatory responses in Alzheimer's disease

It is widely accepted that chronic neuroinflammation plays a role in the development of AD, although the specific mechanisms remain elusive. Microglia are specialised immune cells of myeloid lineage that reside chiefly in the central nervous system and comprise up to 15% of all cell types found in the brain [84]. Their main function is surveillance and maintenance of the central nervous system through clearance of dead and dying cells, as well as plaques [85]. Microglia operate in relative isolation from other immune cells and are highly attuned to pathogenic invasion. They are responsible for recognition, phagocytosis, antigen-presentation and memory in the event of infection [86]. Microglia express the nicotinamide adenine dinucleotide phosphate (NADPH) oxidase (NOX), an enzyme that produces superoxide (O₂ ^.-) and results in the formation of a range of oxidant species [87]. Both the NOX2 and NOX4 isoforms are present in microglia, however NOX2 is most heavily expressed and is particularly reactive to damage response signalling [88–90]. Microglia have receptors for Aβ that trigger the NOX-mediated respiratory burst [87,91,92], and the release of pro-inflammatory cytokines [93–95].

Microglia can also express myeloperoxidase (MPO), albeit at low levels under non-pathological conditions [96]. MPO is a heme enzyme that uses H₂O₂ and a chloride ion (Cl^-) to produce hypochlorous acid (HOCl), or chlorine bleach [97]. HOCl reacts readily with amines to produce chloramines, which are less reactive, longer-lived species with higher cellular diffusion rates [98]. Immune cell-derived oxidants differ greatly in their specificity and reactivities and produce a range of radical and non-radical species that can influence a variety of cellular and molecular processes, but can also cause tissue injury [99]. Importantly, oxidative damage is strongly associated with AD pathology, as demonstrated by increased levels of oxidised and nitrated proteins observed in the AD brain [100–106].

In numerous other diseases, innate immune cells are known to adopt polarised phenotypes, with either pro- or anti-inflammatory properties [107,108]. Similarly, by using transcriptomic and immunohistochemical techniques, many studies have observed that microglia function is altered in AD as several different subtypes of microglia have been identified [109–111]. These data have led to the categorisation of microglia into either M1 (pro-inflammatory and neurotoxic) with increased inflammatory cytokine release and oxidant production or M2 (anti-inflammatory and neuroprotective) that release anti-inflammatory cytokines, increase phagocytosis and promote neuronal survival [112–114]. In animal studies, mice exposed to a single dose of peripherally administered lipopolysaccharide (LPS) produced activated microglia (trained), whereas mice who received four successive doses of LPS produced less reactive microglia (tolerant). Trained mice exhibited increases in both Aβ plaque deposition and total Aβ levels compared to control mice. Intriguingly, the mice who had tolerised microglia had a marked decrease in both measures compared to control [115,116], suggesting that acute immune responses are particularly detrimental in AD pathology. Microglia with a ‘hyper-reactive’ immune response have been observed surrounding amyloid plaques in AD brains [117,118]. These microglia are primed and exhibit an excessive inflammatory response characterised by the release of pro-inflammatory cytokines and oxidants when challenged with systemic infection. This suggests that the AD-associated ‘hyper-sensitive’ microglia phenotype may be epigenetically regulated.

In an accelerated ageing mouse model, a single challenge with LPS significantly increased oxidant production by microglia compared to control [118]. This increased reactivity had direct consequences for cognitive function, as mice exposed to a single dose of LPS performed significantly worse compared to controls in a spatial learning task four weeks after the challenge [119]. The continued activation and release of neurotoxic factors by the microglia in response to infection and/or Aß deposits can lead to a vicious cycle of neuronal cell damage and increased neuroinflammation. Subsequent injury signals that perpetuate microglial inflammation is a process termed ‘reactive microgliosis’ [120–122]. On a mechanistic level, this process is mediated by the damage incurred by oxidants to neuronal cell constituents, however, less is known about the longer-term impact of oxidant exposure on the epigenome and transcriptome of the neuron.

Variants in the gene, triggering receptor expressed on myeloid cells-2 (TREM-2), has provided further evidence that links microglia with AD and cognitive decline [123,124]. Jonsson et al. demonstrated that individuals with a missense mutation in TREM2 had increased risk of developing late onset AD [123]. TREM2 is a surface receptor molecule expressed on innate immune cells including microglia. TREM2 is involved in phagocytosis of bacteria, activation of respiratory burst and resolution of inflammation through the clearance of cellular debris [125,126]. Interestingly, individuals with the variant had significantly decreased cognitive scores even in the absence of AD [123]. Recent work has connected TREM2 with APOE [127], which was identified as a ligand for TREM2 where the association with Aß enhances its uptake by microglial phagocytosis. Importantly, pathological variants of TREM2 associated with late onset AD did not effectively bind APOE and phagocytosis of Aß was impaired [127]. There is substantial evidence that microglia play a role in the pathogenesis of AD. However, it remains unclear whether it is the Aβ deposits and subsequent phagocytosis that drive a pro-inflammatory and cytotoxic microglial state or if microglia first become dysfunctional and lose their anti-inflammatory and neurosupportive role, which leads to the accumulation of Aβ.

Whilst microglia and perivascular macrophages are the predominant resident immune cells in the brain under non-pathological conditions, other cell types can infiltrate the brain in AD, and can influence redox status and inflammatory signalling in the microenvironment.

The role of neutrophils in inflammation and blood–brain barrier dysfunction in AD

Neutrophils are our most abundant white blood cell and the first to respond to infection or damage as part of the innate inflammatory response [128]. Neutrophils contain a large amount of MPO in their azurophilic granules, accounting for an estimated 5% of their dry weight, making them the most likely source of MPO in circulation, at least under homeostatic conditions [129]. Upon phagocytosis, neutrophils destroy engulfed bacteria within specialised phagosomes in part due to the oxidative burst [97,130]. Another feature of activated neutrophils is the ability to extrude DNA-histone complexes called neutrophil extracellular traps (NETs) which serve to sequester and kill pathogens [131]. These NETs are decorated with active MPO, increasing the potential for extracellular oxidant generation [132].

Neutrophil function declines with increasing age, particularly in the ability to accurately migrate to sites of infection and inflammation [133]. In healthy aged individuals it has been demonstrated that neutrophils display off-target chemotaxis, increased primary granule and proteinase release and poor pathogen clearance; traits that increase local inflammation and damage to non-pathogenic tissues [133]. While these fundamental processes seem to deteriorate with age, the ability of neutrophils to produce oxidants remains a matter of debate. Several reports have shown that aged neutrophils constitutively produce oxidants [134–136] and exhibit a reduced capacity for phagocytosis [42,137,138], while others have shown that receptor-induced signalling, particularly response to pathogens, is impaired and therefore the neutrophil oxidative burst is reduced [139,140]. However, these discrepancies may be explained by delayed response times in ageing neutrophils, as oxidant measurements taken at later time points show no significant differences between aged and young individuals [139,141]. Furthermore, aged neutrophils do not respond as rapidly to anti-inflammatory signals released from other immune cells, such as IL-10 [136,142]. These data indicate that aged neutrophils can produce oxidants at a similar magnitude to younger individuals, however, they display a delayed response to stimuli and a blunted response to anti-inflammatory signalling. Taken together, aged neutrophils appear able to migrate to and respond at sites of inflammation but the timing and precision of oxidant release is reduced, therefore, the likelihood of off-target tissue damage during infection and inflammation is increased.

Numerous AD studies suggest that there are both systemic and localised inflammatory factors involved in the pathophysiology of AD, many of which involve neutrophil activity [48,143–145]. An elevated circulating blood neutrophil to lymphocyte ratio has been used as a biomarker for increased inflammatory activity in a range of diseases including AD [146]. MPO is significantly increased in AD patient plasma, compared to age-matched controls and shows a positive correlation with circulating Aß levels [147]. More recently, a meta-analysis has pinpointed MPO as one of the most predictive circulating neutrophil inflammatory markers associated with disease progression in AD [48]. On a functional level, several studies have reported an enrichment of neutrophils with altered phenotypes in AD patients [148,149]. Much like plaque-associated microglia, these neutrophils display a ‘hyper-activated’ phenotype characterised by constitutive superoxide and cytokine production [143]. The abundance of these neutrophils was significantly correlated with Aβ deposition in the brain, as well as rate of cognitive deterioration [143]. Likewise, hyper-activated neutrophils have been reported in a separate study showing an increased basal expression of the adhesion molecule CD11b (a marker of migratory activation) in AD patients compared to controls, with similar associations made with disease severity and mental decline [144]. In contrast, one study has reported that there is no difference in oxidative burst in circulating neutrophils from AD patients and healthy controls, but concluded that the neutrophils themselves are under oxidative stress due to mitochondrial dysfunction [150]. More recently, it was shown that peripheral blood neutrophils from AD patients demonstrated higher oxidative stress evidenced by increased oxidised glutathione to glutathione ratios and malondialdehyde content compared to age matched controls [145]. These AD patients also had higher levels of circulating IL-6 and TNF-α, two pro-inflammatory cytokines with potent chemotactic and stimulatory effects on neutrophils [145]. Although the source of the increased oxidants may be unclear, it is evident that neutrophils in AD patients are under increased oxidative stress which is likely to have an impact on their fate and function.

Under homeostatic conditions, neutrophils are rarely found in the central nervous system, as their passage into the brain parenchyma is restricted by the highly selective endothelial cells that make up the blood brain barrier (BBB). Under conditions of trauma, infection and neurodegenerative disease the BBB becomes disrupted and neutrophils can gain access to the brain parenchyma [151,152]. Disruptions to the BBB are known to be an early biomarker of non-demented cognitive decline [153] and neutrophil-derived oxidants and proteases can modulate barrier properties [143,154]. For example, hypothiocyanous acid (HOSCN), formed through the oxidation of thiocyanate by H₂O₂ catalysed by neutrophil MPO, is a longer-lived oxidant with high affinity for thiols and a likely product at inflammatory sites [155,156]. Sublethal doses of HOSCN have been associated with disrupted brain endothelial cell morphology and function and increased gap formation, suggesting that HOSCN may promote immune cell migration into the BBB [154]. Neutrophil recruitment to the brain may contribute to the reduced cerebral blood flow observed in patients with AD, as depletion of neutrophils in AD mouse models not only improves blood flow but also memory and cognitive function [157,158]. While neutrophils are recruited to the BBB in AD, infiltration into the brain parenchyma is dependent on other immune cells that both summon, and act as gatekeepers to the central nervous system (Figure 1).

Figure 1.

Neutrophil and microglial contributions to oxidative stress and neuroinflammation in Alzheimer’s disease. Under homeostatic conditions, brain-resident microglia have neuroprotective functions that include the clearance of plaques and defective cells through phagocytosis and the release of survival factors. During neuroinflammation neutrophils are recruited to the brain vasculature by the release of cytokines such as IL-6 and IL-8 by reactive microglia. M1 microglia may contribute to the deposition of Aß plaques through frustrated or inefficient phagocytosis. The presence of plaques and tangles can cause the microglia to become activated and release oxidants that can damage nearby neurons. The damage response molecular patterns (DAMPS) released from the damaged neurons cause further activation and oxidant release leading to a vicious cycle of ‘reactive microgliosis’. Neutrophils that enter the brain parenchyma can be stimulated by the release of MIF, IL-2 and DAMPs from the local microenvironment. Neutrophils can also release neutrophil extracellular traps (NETs) that contain the enzyme myeloperoxidase (MPO) that converts hydrogen peroxide (H₂O₂) to hypochlorous acid (HOCl). These oxidants can increase damage and recruit more activated neutrophils, perpetuating the localised inflammatory response. HOCl can also react with amines to form chloramines which are longer-lived species that can freely diffuse into other cells and modify cellular function. Figure created with biorender.com.

Microglia-neutrophil crosstalk has been well documented in the context of LPS-induced neuroinflammation. Direct administration of LPS to mouse brains causes a large influx of neutrophils to the brain, mediated by direct activation of the endothelial cells at the BBB by activated microglia [159]. In contrast, only very high doses of LPS administered systemically recruited neutrophils into the brain parenchyma, and microglia-derived inflammatory proteins were required for infiltration [159]. Once recruited to the brain during induced neuroinflammation, neutrophils have been shown to modulate microglial activation states and even return to systemic circulation after active migration within the brain parenchyma [160]. In humans, using an ex-vivo microfluidic device, it has been shown that Aß-activated microglia recruit neutrophils predominantly through the production of chemoattractant cytokines IL-8 and IL-6 [161]. Neutrophil-microglia interactions resulted in the production of IL-2 and macrophage inhibitory factor (MIF), two cytokines with known associations with AD [162–164]. MIF also acts directly on neutrophils, prolonging lifespan and increasing the production of HOCl, and extracellular traps [165].

Neutrophil infiltration and oxidative stress in the brain parenchyma in AD

Increasing reports indicate that neutrophils are present in the AD brain, however, the extent of their infiltration and their exact role is yet to be fully determined. While neutrophil accumulation in the vasculature is abundantly described in AD, the scarcity of neutrophils within the brain tissue has also been noted [166,167]. However, neutrophil extravasation and migration to amyloid plaques was observed in mouse models using 2-photon in vivo imaging [168]. These images showed that neutrophils accumulate at sites of Aß plaque deposits, but their specific activity was not demonstrated. More recently, positron emission tomography (PET) imaging has been used to visualise neutrophils in mouse brains, and showed that neutrophil infiltration is increased in AD and their presence is associated with microglial activation via granule protein CAP37 [169]. Early reports indicate that Aß fibrils stimulate NADPH oxidase activity in both microglia and neutrophils leading to the release of oxidants, suggesting that Aß can induce oxidative stress in the brain by stimulating immune cells [87,91,170,171].

Depending on the stimuli, the release of neutrophil extracellular traps (NETs) is a process that relies on the formation of oxidants by the NADPH oxidase and MPO enabling more oxidants to be released and produced in the extracellular space [172]. Whilst it has been shown that human neutrophils do form NETs in vitro in response to amyloids such as α-synuclein, Sup35, and transthyretin [173], a recent report has shown that Aß aggregates do not induce NET release in non-inflammatory conditions using neutrophils from healthy controls [167]. Nevertheless, there is compelling evidence placing increased NET activity and MPO in the brains of patients with AD. Zenaro et al. showed that there was an increased number of MPO-positive cells colocalised with typical NET markers, in AD patients’ brains and surrounding vasculature compared to age-matched controls. This was also observed in an AD mouse model where it was demonstrated that neutrophil depletion led to improved memory and cognitive function in those mice [174]. This finding has since been corroborated in a study showing that MPO knockout in a mouse model of AD exhibited significantly reduced neuroinflammation and cognitive decline [175].

Although other sources of MPO in AD have been proposed [176–178], mouse models have demonstrated that neutrophil abundance was positively correlated with MPO levels, with at least 97% of cells positive for MPO co-labelling for the neutrophil marker calprotectin (S100A8) [167]. This would suggest that neutrophil accumulation is the major source of MPO in AD. However, in contrast to previous literature, the authors demonstrated that the majority of NET-forming neutrophils remained confined to the vasculature, and were not associated with MPO deposits around amyloid plaques [167,174]. Rather, MPO associated with plaques within the brain localised with microglial markers. This implies that the neutrophil contribution to oxidative stress in AD may be most significant in connection with the disruption of the vasculature and the BBB. This aligns with the finding from Zenaro et al. showing that neutrophils that migrated to the brain in AD mice produce IL-17, a cytokine that is not only damaging to the endothelial cells of the BBB but also primes and recruits neutrophils [174,179]. Furthermore, neutrophil granule protein CAP37 has been found in the endothelial cells of the brain microvasculature in AD and contributes to the activation of microglia and the release of neurotoxic oxidants into the brain [169,180,181]. Additionally, neutrophils may release microvesicles containing microRNAs that target endothelial cell permeability at the BBB [182]. Furthermore, recent work has demonstrated that neuroinflammatory conditions can facilitate the trafficking of myeloid cells across the blood brain barrier at specific entry points [183]. Regardless of the source of MPO in the brain, it is clear that it contributes to localised oxidative stress as HOCl has been detected in proximity to plaques [184].

Given the sensitivity of the brain to oxidative stress, it could be that only small numbers of neutrophils are required to penetrate the brain parenchyma in order to exert deleterious effects. Or, as we will discuss in the following sections, immune-derived oxidants produce longer-lived and less reactive secondary species, exerting non-cytotoxic effects on neighbouring cells through epigenetic reprogramming.

DNA methylation and Alzheimer’s disease

The clinical presentation of AD is highly heterogenous, even differing between people with identical genetic backgrounds [185–187]. This led to the hypothesis that epigenetic factors may be a strong underlying element in its pathophysiology. DNA methylation is a chemical modification to DNA that allows dynamic regulation of gene transcription in response to environmental stimuli, without constituting a change in the underlying genetic sequence. DNA methylation is the most commonly investigated epigenetic modification and is an important process in regulating gene expression and directing cell function [188–190]. DNA methylation can direct transcription through chromatin structure by influencing accessibility to the DNA, or it can promote or inhibit protein binding to gene elements. The pattern of methylation on genomic DNA is largely established during development, however, certain genomic regions are susceptible to dynamic change, and there is a predicable shift in methylation levels with ageing [191–194]. While localised DNA methylation marks are dynamic, general conservation in global epigenetic patterns is required to maintain cell type fidelity.

DNA methylation patterns in human cells are maintained across cell divisions by DNA methyltransferases (DNMTs). Specialised DNMTs catalyse the transfer of a methyl group from the S-adenosylmethionine (SAM) to the C5 atom of a cytosine – guanine dinucleotide (CpG), in hemimethylated DNA [195]. Therefore, optimal cellular SAM levels are essential to preserve methylation patterns. CpG dinucleotides are often located in proximity to promoter regions, and methylation at these regions generally results in localised gene silencing by inhibiting binding of transcription factors [196].

Although DNMT-driven transfer of methylation is the fundamental mechanism that maintains DNA methylation patterns during cell replication, an alternative mechanism that results in decreased cytosine methylation can occur via the ten-eleven-translocation-1 (TET1) pathway. TET1 is a specific oxidative enzyme, which catalyses the conversion of methylated cytosine to 5-hydroxymethylcytosine (5hmC) and its activity is not restricted to periods of cell replication [197,198]. It has been proposed that active TET mediated demethylation contributes towards the early stages of disease progression [199] where decreased levels of TET enzymes are responsible for selective loss of 5hmC found in the AD brain [200]. In addition, DNMT1, that replicates DNA methylation patterns during cell division, does not recognize 5hmC and this leads to passive demethylation of DNA.

AD pathology has been associated with dysregulated gene expression in the aged and many of the affected genes have been linked to altered DNA methylation. A gradual increase in DNA methylation with age is observed at several sites across the genome [201]. In the past decade, numerous epigenetic clocks have been identified that are able to predict the biological age of a person based on the DNA methylation profiles at specific genes [191–194]. Many of these clocks suggest that accelerated ageing occurs in the AD brain [202]. More recently, brain-specific DNA methylation clocks such as PCBrainAge have been developed that can predict changes leading to AD [203,204]. With ongoing research, several hundred differentially methylated CpGs from various brain regions have been identified and associated with AD, with the most recent study identifying 118 sites in the hippocampus [205]. Because the patterns of methylation are highly conserved within specific brain regions there has been a significant difficulty in attributing pathophysiology. However, there is a consensus that decreased methylation appears to be associated with disease progression. As is often the case with complex disease, individual pathways or genes are not particularly strong risk factors in isolation [192–194]. It is also likely that altered DNA methylation patterns are correlative rather than causative or contribute more to the timing and progression of disease onset [195].

The question remains regarding how the environmental or biological stimuli can trigger the altered methylation patterns observed in AD, and the underlying mechanisms through which they become established within a proportion of cell populations. This is of particular interest in brain tissue, where neurons undergo little or no mitosis [206–208]. It also remains unclear whether the active methylation and demethylation of cytosine in neurons resembles the traditional view of methylation as an epigenetic element [209,210].

Mechanisms of redox regulation of DNA methylation changes in Alzheimer’s disease

It is clear that an environment of oxidative stress contributes to the pathology of AD [211], with oxidative damage being one of the earliest hallmarks [212]. Superoxide anions (O₂•-), hydrogen peroxide (H₂O₂), and hydroxyl radicals (•OH), can cause significant cellular damage, including modifications to DNA, proteins, and lipids. A common consequence of increased oxidative stress is increased DNA damage and failure to adequately repair DNA damage, which is prevalent in AD [213–217]. DNMTs are generally up-regulated in response to DNA damage and also have a high affinity for binding at DNA lesions [218]. This implies that the increased DNA methylation and up-regulation of DNMT1 that has been observed in late onset AD may be a consequence of DNA damage [219].

DNA methylation and oxidative damage

Hypermethylated genes are reported to be vulnerable to oxidative damage mediated by Aβ. Cytosines at CpG sites are susceptible to DNA methylation, while guanine residues are prone to oxidative damage, including formation of 8-oxo-2’-deoxyguanosine (8-oxo-dG), a widely used biomarker of oxidative damage to DNA [220]. Endogenously produced oxidants generate approximately 10⁵ 8-oxo-dG per day [220,221]. These are recognised and repaired by oxoguanosine DNA glycosylase 1 (Ogg1) enzymes, which are reportedly deficient in AD and other ageing-related diseases [220]. Both 5-methylcytosine and 8-oxo-dG have demonstrated suppressive activity on transcription machinery. This effect is greater when both 5-methylcytosine and 8-oxo-dG are present at a CpG site as the ability of Ogg1 to repair 8-oxo-dG is repressed when it is preceded by 5-methylcytosine [222,223]. Increased levels of Aβ promote oxidant generation, which leads to DNA damage and accelerated neurodegeneration [220] (Figure 2).

Figure 2.

Potential mechanisms of redox regulation of epigenetic pathways in Alzheimer’s disease. Neuroinflammation leads to increased oxidative stress in the brain. Oxidants produced by immune cells have different reactivities and diffusion rates. Hydrogen peroxide (H₂O₂) has a high affinity for thiols and DNA methyltransferases (DNMT) have a key cysteine residue in their active site. Chloramines produced from the reaction of amines with hypochlorous acid (HOCl), will readily react with methionine which can reduce the availability of S-adenosyl methionine (SAM); a major methyl donor. Inhibition of DNMT through its active site and depletion of SAM can impede the DNA methylation and lead to hypomethylation and aberrant gene transcription. Conversely, DNA damage can stimulate DNMT to bind to DNA and cause hypermethylation around the strand break which can impair repair processes and the upregulation of genes required by the cell. Created with biorender.com

Paradoxically, it has been shown that oxidative stress can also result in the downregulation of the DNMT1 and DNMT3 enzymes and corresponds with increased hypomethylation in the promoter regions of the APP, BACE1 and PSEN1 genes [224–227]. Disrupted cleavage of the amyloid precursor protein (APP) is critical to the overproduction of Aβ. Proteolytic cleavage of APP is carried out by the β-site APP-cleaving enzyme 1 (BACE1) and the ỿ-secretase complex in which PSEN1 embodies the catalytic enzyme. All three genes are regulated by DNA methylation in their promoter regions. Notably, the promoter region of the APP gene, with a CpG content of around 72%, has been observed to be completely demethylated in AD postmortem studies [224]. In a separate study, the treatment of human neuroblastoma cells with H₂O₂ led to changes in BACE1 expression [228]. In response to cellular stress, phosphorylated (activated) eukaryotic transition factor-2-alpha (eIF2α) stops protein translation, however, BACE1 and other stress response genes become more highly expressed [228]. Elevated levels of phosphorylated eIF2α observed in AD brains suggest a pathway through which redox signalling may contribute to AD pathology through influencing the translation of BACE1 protein expression [228].

An under-researched and novel mechanism that links oxidative stress and DNA methylation to AD is by its contribution towards circadian deregulation [229]. In mammalian cells, the core components of the biological clock include the positive regulators BMAL1, CLOCK, and NPAS2, and the negative regulators CRY1/2 and PER1/2/3 [230,231]. Circadian rhythms are generated by the transcriptional/translational oscillation of these core components, which subsequently regulate >10% of the transcriptome [232]. This facilitates metabolic and physiological functions in synchronizing with predictable changes in the environment. The circadian rhythm regulates production of oxidants, scavenging, and transcription of oxidant-responsive genes [233]. Disruption of circadian rhythms leads to increased oxidative stress in neurons [234], neuroinflammation and oxidative stress in the neural tissue. Sleep deprivation and disruption to circadian rhythm interact and increase the risk for the development of AD [235], can modulate the epigenetic landscape [236], and increase oxidative stress and inflammation [237].

Immune-derived oxidants and epigenetic regulation

The effects of DNA damage on gene expression in AD have been clearly described, however the impact on redox signalling pathways is less clear. It is known that the key molecular drivers of maintenance DNA methylation can be regulated by oxidants, with the resulting altered methylation patterns correlating with subsequent gene expression, and may have important implications for the pathogenesis of AD. As described earlier, a variety of oxidants are produced by microglia, neutrophils and macrophages. These species are able to react with cysteine and methionine within the cell [28]. DNMTs contain a cysteine residue in their active site, which is proposed to be a primary target for oxidative modification [238]. SAM serves as the sole methyl donor in several methyltransferase reactions, therefore decreasing the availability of free methionine also decreases SAM availability for methylation [239]. This demonstrates the potential of immune derived oxidants to alter epigenetic pathways by direct modification of DNMT or indirectly, by interfering with one-carbon metabolism through methionine oxidation. We have demonstrated, through in vitro studies of cell lines, that there is a relationship between oxidative stress and epigenetic modifications, with results suggesting this pathway could be of relevance in diseases such as AD [32,239,240].

Reduced levels of SAM in the cerebrospinal fluid of AD patients has been reported. The autopsied brain samples of patients with AD and neurologically normal controls demonstrated that mean levels of SAM and S-adenosyl-L-homocysteine (SAH) were significantly reduced in all five different regions of the AD brain including the frontal cortex, occipital cortex, temporal cortex, putamen, and hippocampus [241,242]. This has led to extensive debate regarding the efficacy of using dietary SAM supplementation in the treatment of AD. However, it has been reported that excess levels of SAM disrupt normal biological rhythms and is metabolised to adenine and methylthioadenine, two potent inhibitors of methylation [243].

Taken together, this provides further correlative support for a linked interaction between oxidative stress and DNA methylation in AD. Understanding these interactions in more detail will provide a clearer picture of the molecular mechanisms driving AD pathology and provide potential therapeutic targets for intervention.

Potential therapeutic strategies targeting inflammation, oxidative stress and epigenetic processes in AD

While novel approaches such as the anti-amyloid antibody lecanemab [244] represent exciting advancements in the treatment of AD, it is increasingly clear that addressing amyloid alone may not be sufficient to produce lasting therapeutic effects. To achieve long-term benefits, comprehensive strategies should target various aspects of the disease, including inflammation, oxidative stress and epigenetic dysregulation. This section explores existing and potential therapies addressing these underlying features of AD.

Targeting DNA methylation in AD

Given the evidence that aberrant DNA methylation is involved in AD pathogenesis, hypomethylating agents could help restore gene expression and ameliorate neurodegenerative processes. Hypomethylating agents such as 5-azacytidine and decitabine (5-aza-2'-deoxycytidine) are approved for use in the treatment of haematological malignancies and could have application in AD [245]. These agents incorporate into DNA and degrade DNMTs, leading to widespread hypomethylation. However, the non-specific and cytotoxic nature of these agents carry the risk of contributing to increased neuronal death and inflammation in the brain. The development of targeted treatments, such as CRISPR-dCas9-based editing systems could offer a precise method for altering DNA methylation at specific gene loci [246]. Once optimised, this methodology combined with cell type-specific and blood–brain-barrier-penetrating technologies could prove a valuable tool in AD treatment [247]. It is worth noting that key genes involved in Aβ processing (discussed previously) are often hypomethylated in the AD brain suggesting that existing conditions contribute to an unfavourable methylation state at these loci. As the field awaits advances in precision epigenetic therapeutics, modifying lifestyle and environmental factors that are known to be associated with detrimental epigenetic patterns represents a holistic approach to prevention of AD.

Targeting neuroinflammation in AD

A number of existing anti-inflammatory agents are currently being investigated for their potential to reduce neuroinflammation AD [248]. One emerging target is histamine, a biogenic amine that functions as both a neurotransmitter and an immune system mediator [249,250]. Although alterations in histamine levels in AD have long been recognised, research into their precise role has progressed slowly [251,252]. In the brain, histamine is primarily produced by neurons located in the tuberomammillary nucleus of the hypothalamus and influences various cognitive functions, including memory and learning. Histamine, histamine-releasing factor and receptor binding levels are dysregulated in AD with low levels corresponding with decreased cognitive ability [253–256]. Our group has demonstrated that histamine reacts rapidly with HOCl and that the resultant histamine chloramines are cell permeable, longer lived and more selective for thiols than HOCl [257]. Since it has been demonstrated that MPO is present in AD brains, it is conceivable that HOCl could contribute to the low histamine levels observed in AD patients. Moreover, due to histamine chloramine’s potent reactivity with thiols and methionine, it also has the potential to interfere with DNMT activity; and alter methionine levels, modulating DNA methylation [31,239].

Histamine interacts with microglia, and is capable of both promoting and suppressing inflammation depending on the context, such as the presence of other inflammatory signals [258]. While histamine and lipopolysaccharide individually increase microglial oxidant production and phagocytosis, their combined exposure significantly reduces these effects. This suggests that increasing histamine levels in AD has potential to mitigate inflammation-induced oxidative stress and increase neuronal activity [258,259].

The use of histamine as a therapeutic has shown promise in pre-clinical models [260] but is complicated by the fact that histamine receptors are not unique to microglia but are also expressed by neurons and astrocytes [261]. As these receptors differ in structure and brain region, combinations of existing histamine receptor agonist and antagonists may be required to ensure targeted treatment [262].

Neuroendocrine disruptions, particularly central insulin resistance, are increasingly recognised as key contributors to AD pathology [263,264]. This resistance impairs glucose uptake and utilisation in the brain, leading to altered signal transduction and contributing to the histopathological changes and hypothalamic inflammation observed in AD [265]. Interestingly, anti-diabetic drugs have shown potential benefits in AD. Notably, intranasally administered insulin has emerged as a prominent treatment, improving memory and plasma Aβ levels in patients with mild cognitive impairment and AD [266].

Pertinently, an extract from the petals of the Dahlia pinnata Cav. (Asteraceae) flower, rich in the flavonoids: butein, sulfuretin and isoliquiritigenin, has been shown to improve glucose tolerance and reverse markers of hypothalamic inflammation in mice and zebrafish and is approved for use in humans [267,268]. In separate studies, butein treatment reduced oxidative stress in neurons and suppressed microglial inflammation in mice [269,270]. Although Dahlia extract treatment has not yet been examined directly in the context of AD, it is a non-toxic, over-the-counter compound that warrants further investigation.

Targeting systemic inflammation in AD

Systemic inflammation plays a critical role in the progression of Alzheimer’s disease (AD), but most nonsteroidal anti-inflammatory drugs (NSAIDs) have shown limited efficacy in treating the condition [271,272]. However, diclofenac, a commonly used NSAID, has emerged as a potential therapeutic for AD due to its unique mechanism of action. Unlike other NSAIDs, diclofenac targets multiple pathways, including the inhibition of the NLRP3 inflammasome, a key contributor to neuroinflammation and a central driver of AD pathology [273,274]. Diclofenac has been associated with a reduced incidence of AD and slower cognitive decline, which may be due to its ability to curb neuroinflammation through inhibition of the NLRP3 pathway [271,272]. The NLRP3 inflammasome is activated by components released from dysfunctional mitochondria, which leads to the secretion of pro-inflammatory cytokines like interleukin-1β and the release of mitochondrial DNA, leading to amplified inflammation and neuronal damage [73]. Despite these promising effects, diclofenac has also been shown to induce mitochondrial dysfunction and contribute to hepatocellular injury by increasing mitochondrial H₂O₂ production [275]. This highlights the potential need for combining diclofenac with mitochondria-targeted antioxidant therapies to mitigate these effects. Such combination therapies could reduce oxidative stress while still harnessing diclofenac’s potent anti-inflammatory properties.

Several mitochondria-targeted antioxidant therapies are available that could help reduce and prevent oxidative damage in affected tissues. One promising compound is mitochondria-targeted quinone (MitoQ), a compound that accumulates within mitochondrial membranes and scavenges excess oxidants [276]. MitoQ can cross the blood–brain barrier, improve cognitive function, and reduce amyloid plaque burden in mouse models of AD, making it a strong candidate for further investigation in AD treatment [277–279].

Similarly, ascorbate (vitamin C) offers significant potential in addressing oxidative stress, inflammation and epigenetic regulation in AD [280]. As well as being a potent antioxidant, ascorbate chelates metal ions like iron, copper, and zinc, which are known to contribute to Aβ aggregation and neurotoxicity [281]. Several studies suggest ascorbate treatment reduces amyloid plaque burden and cognitive deficits in animal models and attenuates amyloid oligomerisation [282–284].

In addition to its well-established antioxidant properties, ascorbate plays a crucial role in the regulation of epigenetic processes, particularly through its influence on TET enzyme activity [285]. Ascorbate serves as an essential cofactor for TET enzymes, enhancing their activity and promoting DNA demethylation, which may help restore normal gene expression profiles that are altered in AD [286]. By promoting TET-mediated demethylation, ascorbate could help reverse these epigenetic changes, potentially restoring the expression of critical neuroprotective genes. Although the use of ascorbate in human AD studies have produced mixed results, some studies have demonstrated a correlation between higher ascorbate intake and reduced AD risk and higher cognitive health scores [287,288]. The wide-ranging biological effects and low cytotoxicity of ascorbate make it a promising compound for further research in AD.

The integration of these therapeutic strategies targeting redox balance, inflammation and epigenetic pathways holds promise for developing effective treatments for AD. Future research should focus on clinical trials to validate the efficacy of these therapies and explore their potential synergistic effects when used in combination.

Summary and future perspectives

Age-related illnesses and disabilities represent an ever-increasing source of strain on healthcare systems worldwide. Ageing continues to be one of the strongest risk factors for late onset Alzheimer’s disease and screening and treatment options are limited. Chronic low-grade inflammation is a characteristic of ageing and systemic inflammation is associated with AD onset, and we have presented a multitude of studies that suggest an effector role for immune cells in AD pathology. The extent to which peripheral immune cells, such as neutrophils, can enter the brain remains unclear and is difficult to measure temporally, however signs of oxidative stress are evident and clearly contribute to the aetiology of AD. Sources of oxidative stress are abundant in AD and include dysfunctional mitochondria [62,289], neurons [290,291] and endothelial cells [292,293], but immune cells are emerging as an abundant and potentially modifiable source. Exciting recent discoveries uncovering the sites of immune infiltration into the brain, will be invaluable in the development of novel therapeutic targets in neurological disease [183].

Oxidative stress can alter neuronal health both by directly damaging the DNA and causing cell death but also in more subtle ways, through the manipulation of key cellular enzymes and cofactors that have the potential to modify the epigenetic regulation of the genes associated with Alzheimer’s disease onset and progression. Further studies are required to explore the impact of immune-derived oxidants on DNA methylation profiles in the ageing brain with the aim of uncovering targeted immunomodulatory, epigenetic or mitochondrial therapeutic agents in the treatment of AD. As the world’s population ages, it will become increasingly important to find reliable biomarkers of oxidative stress in middle-aged humans, before the onset of age-related disease such as AD, with the ultimate goal of prolonging the health span of individuals as they age.

Funding Statement

This work was supported by Canterbury Medical Research Foundation; Health Research Council of New Zealand; Wellington Medical Research Foundation.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Data availability statement

All data is publicly available.