Molecular mechanism of Alzheimer’s disease using integrated multi-omics

Based on the age of onset of the first symptoms, AD can be categorized as Early (also known as familial) and Late (also known as sporadic) -onset AD (EOAD and LAOD), where EOAD affects individuals under the age of 65 and accounts for 1–5% of all AD cases (Dorszewska et al., 2016). In contrast, LOAD effects population of ages above 65 years of age (Mendez, 2017). The two subtypes present differences beyond the onset of symptoms, including clinical manifestations, neuropathology and neuropsychological profiles. It has been reported that the main genetic causes of EOAD are identified as mutations in genes such APP on chromosome 21, presenilin 1 (PSEN1) on chromosome 14, and presenilin 2 (PSEN2) on chromosome 1 (Levy-Lahad et al., 1995).

Risk alleles are genetic variants that increase susceptibility to developing certain diseases. In the context of AD, identifying these alleles is crucial for understanding the underlying genetic mechanisms that contribute to the pathophysiology of this disorder.

To identify risk alleles associated with AD, a comprehensive search was conducted in the GWAS Catalog1, a curated resource of GWAS findings. This research focused on genetic variants linked to AD (Table 1), extracting relevant alleles associated with disease risk. The identified risk alleles were then sorted by their p-values to prioritize those with the strongest statistical significance. From this analysis, the top 10 genes with the lowest p-values were selected, representing the most robust genetic associations.

Among the top genetic loci associated with AD, the APOE gene stands out as the most significant and well-established risk factor, particularly the ε4 allele, which markedly increases susceptibility to late-onset AD. APOE plays a critical role in lipid metabolism, amyloid-β aggregation, and neuroinflammatory processes as discussed earlier. Recent studies highlight that its effect may be modulated by neighboring variants, including TOMM40 and APOC1. Polygenic profiles combining these variants have shown distinct associations with cerebrospinal fluid (CSF) and plasma biomarkers, particularly Aβ42 and tau (Kulminski et al., 2022).

These nearby genes are frequently co-inherited with APOE due to their close genomic proximity and may contribute independently to AD risk. APOC1 has been implicated in modulating inflammatory responses and cholesterol transport, while TOMM40 is involved in mitochondrial protein import, with certain polymorphisms influencing the age of AD onset and promoting neuroinflammatory pathways (Kulminski et al., 2022; Zhang et al., 2024; Chen et al., 2023).

Although the APOE gene cluster is often described as a tightly linked genomic region, emerging transcriptomic evidence suggests a more nuanced regulatory architecture within this locus. Integrative analyses of postmortem human brains have demonstrated that while APOE and APOC1 mRNA levels closely track overall cluster expression, TOMM40 exhibits distinct age- and disease-dependent transcriptional behavior, with reduced expression in younger AD cases independent of global APOE cluster patterns (Haghani et al., n.d.). Proteome-wide association studies integrating AD GWAS data have demonstrated that transcriptomic and proteomic associations show only partial concordance, with aligned mRNA and protein-level effects observed for a limited subset of loci (Wingo et al., 2021). Notably, the genes discussed here do not fall within this concordant subset identified in large-scale ROSMAP-based analyses, highlighting the importance of post-transcriptional regulation and multi-layered molecular control in AD (Wingo et al., 2021).

Taken together, these findings suggest that the chromosome 19 risk locus is regulated in a more complex manner than often assumed, with shared genetic risk existing alongside gene-specific transcriptional and proteomic effects. These patterns highlight the value of multi-omics approaches for understanding how closely linked genes contribute to AD.

Another gene within the chromosome 19 risk region, NECTIN2 (PVRL2), has emerged as an independent contributor to Alzheimer’s disease susceptibility. Genetic association studies have consistently implicated variants at this locus, particularly rs6859, in AD risk (Kunkle et al., 2019; Lambert et al., 2013). Recent mediation analyses further suggest that the effect of rs6859 on disease risk is partially mediated through elevated cerebrospinal fluid pTau-181 levels, linking NECTIN2 variation to tau-related neurodegenerative processes (Rajendrakumar et al., 2024a). In longitudinal analyses, NECTIN2 variants have been associated with differential trajectories of cognitive decline, suggesting that genetic variation at this locus may contribute to inter-individual heterogeneity in disease progression (Rajendrakumar et al., 2024b). Functionally, NECTIN2 encodes a cell-adhesion molecule of the nectin family that plays roles in synaptic organization, intercellular junctions, and immune-cell interactions (Liu et al., 2012). Multi-omic analyses provide additional support for its biological relevance; the NECTIN2 locus exhibits both cis-eQTL effects in human brain tissue and cis-pQTL effects in cerebrospinal fluid, indicating that genetic variation influences both mRNA expression and protein abundance (Wang et al., 2024). Notably, functional validation experiments demonstrate that NECTIN2 overexpression modulates soluble TREM2 levels, suggesting a potential link between this locus and microglial signaling pathways implicated in AD (Wang et al., 2024). Together, these findings indicate that NECTIN2 contributes to Alzheimer’s disease through mechanisms that extend beyond simple linkage with APOE, involving coordinated genetic, transcriptional, and proteomic regulation.

The observation that several of the most significant AD-associated variants converge on genes involved in immune regulation is important to highlight. Importantly, large-scale systems-level analyses have demonstrated that these immune and lipid-associated risk loci do not operate as isolated signals but are embedded within coordinated molecular networks. Network-based integrative analyses integrating genomic and transcriptomic data across multiple brain cohorts have identified AD-associated gene modules enriched for immune response, synaptic signaling, and metabolic pathways. For example, Wang et al. applied a systems biology framework to define co-expression modules that capture the convergence of genetic risk variants on microglial and inflammatory pathways, highlighting regulatory hubs rather than single-gene effects (Wang et al., 2018). Similarly, Wan et al. (2020) employed integrative network modeling approaches to delineate disease-associated transcriptional modules, revealing that AD susceptibility genes cluster within coordinated immune and glial regulatory networks. These findings reinforce the concept that AD pathogenesis reflects network-level dysregulation across molecular layers rather than parallel, independent biological processes (Wan et al., 2020). Notably, these network-level observations are consistent with large-scale GWAS meta-analyses, which have repeatedly identified immune and lipid-processing pathways as core components of the genetic architecture of Alzheimer’s disease. Large-scale GWAS meta-analyses have consistently demonstrated that pathways related to immunity and lipid processing represent central components of the genetic architecture of Alzheimer’s disease, rather than secondary consequences of neurodegeneration (Kunkle et al., 2019; Lambert et al., 2013). Importantly, integrative analyses across genetic and transcriptomic datasets indicate substantial inter-individual heterogeneity in AD, with distinct molecular subtypes characterized by differential immune activation signatures (De Jager et al., 2018). These findings suggest that immune-related loci may contribute to disease susceptibility and to variability in clinical manifestations and disease progression.

In this context, single-nucleus transcriptomic profiling of human prefrontal cortex demonstrates that substantial transcriptional remodeling occurs early in Alzheimer’s disease, prior to severe neurofibrillary pathology or advanced cognitive decline (Mathys et al., 2019). Notably, early gene-expression changes in glial populations are enriched for immune-related regulators. In microglia, genes such as APOE, C1QC, RASGEF1B, and CD83 are differentially expressed, consistent with activation of innate immune and complement pathways. In astrocytes, early upregulation of GFAP, SLC1A2, and CEMIP2 reflects reactive and metabolic reprogramming states (Mathys et al., 2019). These findings indicate that immune and glial transcriptional activation is detectable at early disease stages, supporting the view that neuroinflammatory processes are embedded in the initial molecular frame of Alzheimer’s disease.

Notably, emerging evidence indicates that APOE may interact with hypothalamic–pituitary–adrenal (HPA) axis dysfunction to accelerate AD pathogenesis. The APOE ε4 allele has been specifically linked to elevated CSF cortisol levels in both AD patients and non-demented older adults (Peavy et al., 2007; Peskind et al., 2001), suggesting a genotype-dependent alteration in HPA axis regulation. This genetic predisposition appears to create heightened vulnerability to environmental stressors, as demonstrated by studies showing that ε4 carriers experiencing prolonged stress exhibit significantly worse memory performance and higher cortisol concentrations than non-carriers under similar conditions (Peavy et al., 2007). The interaction between genetic risk and HPA axis dysfunction may be further modified by personality factors, with neuroticism and extraversion amplifying the adverse effects of APOE ε4 on cognitive outcomes (Dar-Nimrod et al., 2012).

Microglia play a direct and vital role in neuromodulation under physiological conditions, whether by physical contact or through multiple receptors and signaling pathways, as mentioned earlier, the complement system acts as “tags” for pruning weak or immature synapses and subsequent engulfing. Microglia not only prune synapses but also interact with engram neurons, the physical representations of memories in the brain, as they are the neurons that are involved in different steps of memory formation (encoding, consolidation, and retrieval) (Guskjolen and Cembrowski, 2023). The inhibition of the complement system using CD55 or depletion of microglia hindered the natural forgetting process and impaired the uncoupling of engram cell networks (Wang et al., 2020). These findings highlight how immune signaling, particularly cytokine production by microglia, contributes to cognitive processes, with IL-1β, TNF-α, and IL-6 standing out as key regulators.

IL-1β was reported to interfere with contextual fear memory formation (Gonzalez et al., 2013), an effect that might be mediated by activating P38/MAPK pathway in dorsal hippocampus. The intrahippocampal injection of IL-1β also delayed ERK phosphorylation and reduced BDNF expression and glutamate release.

TNF-α role in learning and memory was only reported in few studies. Mice lacking TNF-α preformed worse in novel object recognition task and Barnes maze than control (Baune et al., 2008). On the other hand, mice injected intraperitoneally with TNF-α had better performance in avoidance test (Brennan et al., 2004). TNF-α was also found to regulate synaptic plasticity, under physiological conditions, TNFα acts as a key permissive factor for glutamatergic gliotransmission in the hippocampal dentate gyrus. Although astrocytes can still exhibit calcium signaling in its absence, TNFα is required to enable the vesicular release of glutamate, which is essential for modulating synaptic activity (Santello et al., 2011).

IL-6 is a pleiotropic cytokine that plays a dual role in the central nervous system, contributing to both homeostatic and pathological processes. Under physiological conditions, IL-6 is involved in synaptic plasticity, neurodevelopment, and memory formation (Gruol, 2014; Balschun et al., 2004). However, its role is highly context-dependent, with both beneficial and detrimental effects reported depending on its concentration, timing, and cellular origin (Erta et al., 2012; Raison et al., 2006). In the healthy brain, low levels of IL-6 may support neuronal communication and neurogenesis, while sustained or excessive production, especially during neuroinflammation—has been linked to cognitive impairment and neurodegeneration (Marsland et al., 2008; Weaver et al., 2002). These findings underscore a complex, bidirectional role for immune signaling in the brain, balancing the mechanisms that underpin healthy cognitive function against those that may drive mood disorders.

In a chronic IL-6 overexpression model, working memory but not associative learning was compromised as evident by Y-maze and fear conditioning test, respectively, (Marguerite Wagnon et al., 2023). mRNA expression levels for a battery of markers were tested in different areas of the brain, interestingly, the cerebellum showed the greatest differences in transcriptomic profile. IBA-1, TREM2, and NF-ΚB levels were increased, which indicate microglial activation and inflammatory reaction. Moreover, the apoptotic marker Cas3 levels increased while the neuro-marker Ero2 decreased, this may indicate a loss on neuron numbers in the cerebellum.

The brain is not solely influenced by external stimuli and experiences, but is also intricately modulated by the immune system, which responds dynamically to these experiences, This dynamic relationship between immune signaling and neural activity plays a crucial role in processes like neuroplasticity, memory formation, and cognitive function, demonstrating the immune system’s active involvement in shaping both brain structure and behavior.

Components of the innate immune systems such the transcription factor NF-κB—play dual roles in promoting synaptic plasticity and cognitive reserve, while also contributing to neuroinflammation and pathological conditions. NF-κB serves as a central mediator that converts synaptic activity into long-term changes in gene expression essential for memory formation (Oikawa et al., 2012; Albensi and Mattson, 2000). Upon neuronal stimulation—often triggered by synaptic events like calcium influx—NF-κB is activated and translocate from the synapse to the nucleus. There, it coordinates a transcriptional program that involves key genes such as JunD, MHC Class I (and II with beta 2 microglobulin), prodynorphin, BDNF, and the more recently identified IGF2 (Gobin et al., 2001). These genes, many of which also share regulation by CREB, collectively enhance synaptic plasticity by promoting spine formation, regulating receptor composition (including AMPA and NMDA receptors), and even facilitating synaptic adhesion (Schmeisser et al., 2012). In essence, NF-κB functions as a molecular hub that translates transient synaptic signals into durable changes in synaptic structure and function, thereby supporting the maintenance of long-lasting memory.

Additionally, NF-κB intersects with BDNF signaling, a critical pathway for synaptic plasticity, neuronal survival, and cognitive function. In AD, disruptions in BDNF/NF-κB crosstalk can compromise neurotrophic support and exacerbate amyloid-β and tau pathology (Gao et al., 2022). NF-κB activation suppresses BDNF expression, while BDNF deficiency may impair neuronal resilience to inflammatory insults, creating a vicious cycle of neurodegeneration (Jones and Kounatidis, 2017). Moreover, chronic activation of NF-κB in the hippocampus and entorhinal cortex has been shown to impair neurogenesis and synaptic function, reducing the generation and integration of new neurons and contributing to cognitive decline (Mattson and Meffert, 2006). Importantly, pharmacological inhibition of hyperactive NF-κB in animal models restores neuronal excitability and spatial memory, supporting the concept that NF-κB blockade can ameliorate cognitive deficits (Soleimani et al., 2025). Together, these pathways underscore the multifaceted role of NF-κB in mediating neuroinflammation, impaired neurotrophic signaling, and reduced neurogenesis—each of which contributes to the progression and persistence of Alzheimer’s disease.

Studies have shown that physiological factors such as higher education and cognitive activity can protect against AD (Wada et al., 2018; Sando et al., 2008). Subjects with higher education showed reduced odds ratio of developing AD, the effect reported was dose-dependent, meaning that the protective effect of education increased with the increase of years in education (Sando et al., 2008). Cognitive activity, as in participating in cognitively stimulating tasks such as reading a newspaper, playing puzzles and card games has been demonstrated to reduce cognitive decline due to aging by 47%. Moreover, cognitive activities were also shown to reduce the risk of developing AD by 33% in a four and half year’s study (Wilson et al., 2002). Higher education and cognitive activities also appear to protect against depression (Bjelland et al., 2008); a meta-analysis that included 36 studies varying between cross-sectional and longitudinal studies concluded that elderly with lower education were at higher risk of depression with odds ratio of 1:58 (Chang-Quan et al., 2010). Moreover, individuals with higher premorbid intelligence are thought to have a greater buffer that can help mitigate the cognitive commonly associated with depression (Venezia et al., 2018).

A plausible link between previously mentioned domains is most likely brain connectivity and synaptic enhancement. Specifically, higher education is hypothesized to protect against AD by increasing cognitive reserve, defined as the brain adaptability that might explain the discrepancies in cognitive abilities and daily functions in response to aging, pathology, or insult. In other terms, education may mediate the ability of the brain to compensate against cognitive decline by recruiting alternative neural networks developed through years of education. The molecular mechanisms involved in cognitive reserve are poorly understood. A study utilized MRI-based cortical thickness analysis to identify brain regions associated with higher educational levels. Following this, the researchers explored the transcriptional architecture of these regions by utilizing brain-wide regional gene expression data and GSEA analysis (Bartrés-Faz et al., 2019). The GSEA analysis revealed a strong enrichment of gene sets related to neurotransmission and immune response in years of education-related areas. Notably, genes involved in ionotropic neurotransmission, including those coding for receptors for glutamate, GABA, acetylcholine, and serotonin, were predominantly upregulated. This suggests an enhanced role of excitatory and inhibitory signaling in these areas, possibly contributing to the cognitive processes associated with years of education. The identification of neuropeptide signaling, G-protein-coupled signaling, and purinergic pathways highlight the complexity of neurotransmission regulation in these regions. Additionally, the immune response gene sets were notably enriched, featuring genes involved in pathogen recognition (TLR3 and 7), cytokine signaling, and immune activation (TGFB1, AIF1, PYCARD). Findings from literature support the role of TLR7 in memory formation and cognition, activating TLR7 can enhance contextual fear memory, suggesting that TLR7 signaling directly influences memory consolidation (Kubo et al., 2012). Another study shows that TLR7 deletion alters the expression profiles of genes related to neural function particularly synaptic proteins related to neuronal development and differentiation such as GRW2A, DAG67, and GRM5 (Hung et al., 2018a). Recent findings also underscore the critical role of TLR9 in memory formation (Jovasevic et al., 2024). In this study, hippocampal neurons were shown to experience double-stranded DNA breaks following learning, which in turn activated TLR9 signaling. This receptor-mediated cascade engaged several essential processes—DNA repair, centrosome function, and the formation of perineuronal nets—that are crucial for stabilizing and consolidating memory traces. Importantly, disruption of TLR9 function resulted in impaired memory and increased genomic instability, emphasizing the pathway’s significance in maintaining cognitive function. The enrichment of TLR3 and TLR7 in education-related brain regions suggests that, similar to TLR9, they may contribute to memory formation and the establishment of cognitive reserve. It is important to note, however, that these receptors differ in their signaling pathways and brain region specificity. TLR3 generally signals via the TRIF pathway (Hung et al., 2018b), whereas TLR7 predominantly utilizes MyD88 (Fishelevich et al., 2011). Moreover, studies such as Jovasevic et al. have focused on TLR9 in the hippocampus (Jovasevic et al., 2024)—a key area for memory formation—the current findings relate to cortical regions. This suggests that while TLR3 and TLR7 might share a convergent role in supporting cognitive functions, their specific contributions could vary depending on the brain area and the receptor-specific signaling cascades involved. This indicates that immune processes might not only play a protective role but could also influence synaptic functioning and plasticity in these brain areas.

The involvement of immune response in cognitive reserve is further supported by findings from Yegla and Foster (2022), in their study, the researchers identify immune-related pathways as significant contributors to cognitive resilience. Specifically, the study highlights genes such as TREM2, C1qa, and P2RX7 that play crucial roles in microglial activation and neuroinflammation. These genes were found to be enriched in cognitive reserve-related regions, emphasizing the potential link between immune function and cognitive resilience. The overlap in genetic regulation between cognitive reserve and AD suggests that the immune mechanisms contributing to cognitive resilience may also influence AD pathology. These findings imply that individuals with a genetic profile favoring a robust, well-regulated immune response could potentially counteract or delay the detrimental effects of AD-related neuroinflammation. In essence, the same regulatory pathways that bolster cognitive reserve might provide neuroprotection, offering promising targets for therapeutic interventions that not only enhance cognitive resilience but also mitigate the progression of AD.

Environment enrichment (EE) studies in rodents may closely model cognitive reserve (Nithianantharajah and Hannan, 2009), the concept of EE is allowing rodents to have enriched sensory and cognitive and motor stimulation compared to normal housing conditions. Many EE studies have shown positive physical and mental effects on rodents, including enhanced cognitive function and reduced depressive symptom (Brenes et al., 2020; Huang et al., 2021; Cordner et al., 2021; Wei et al., 2020). EE was found to delay cognitive impairments in AD mouse model, AD mice treated with EE showed increased synaptic density and reduced neural inflammation.

Interestingly, serum levels of IFN-γ were elevated in EE treated mice which in turn enhanced the secretion of exosomal micro-RNA 146a (miR-146a) from choroid plexus cells and possibly mediated the cognitive rescue (Nakano et al., 2020). miR-146a acts as a negative feedback regulator of astrocytes inflammation by inhabiting NF-κB (Iyer et al., 2012). Similarly, EE was found to enhance cognitive function examined by spatial learning via the Morris water maze, those mice were found to have more complex dentate gyrus of the hippocampus. This enhancement was accompanied by an increase in protein levels of brain derived neurotrophic factor (BDNF), phosphorylated cyclic adenosine monophosphate (cAMP) response element-binding protein (pCREB), both reported to be important regulators of hippocampal neurogenesis and neuroplasticity. CREB is a transcription factor that induces and regulates many genes related to neurotransmitters, growth and transcriptional factors, and metabolic enzymes. CREB involvement in neural plasticity and memory is vastly proven by studies on CREB knockout mice that presented cognitive dysfunctions shown in fear conditioning tests. Moreover, mice that constitutively express CREB were reported to have a lower threshold for induction of late phase long term potentiation (Sakamoto et al., 2011).

Recent evidence has introduced the provocative hypothesis that AD may, in part, be driven by autoimmune processes. Traditionally, chronic neuroinflammation has been recognized as a feature of AD; however, emerging data suggest that dysregulation in the brain’s immune surveillance may lead to an autoimmune reaction against neuronal components. This autoimmunity could contribute to the progressive synaptic loss and neurodegeneration characteristic of AD. The proposed autoimmune aspect of AD raises intriguing questions about the underlying mechanisms, suggesting that it might arise from, or be exacerbated by, maladaptive immune responses—albeit at different points along a spectrum of immune dysfunction.

Genetic studies have revealed robust associations between AD risk and polymorphisms in major histocompatibility complex (MHC) genes, including the human leukocyte antigen (HLA) system (Kunkle et al., 2019; Marioni et al., 2018; Misra et al., 2018; Jones et al., 2015), which are well recognized for their roles in autoimmune disorders such as multiple sclerosis and systemic lupus erythematosus (Alcina et al., 2012; Furukawa et al., 2014). In addition, large-scale genomic analyses have identified overlapping risk variants in genes like CLU, CR1, and BIN1 that connect AD pathology with autoimmune-mediated processes (Yokoyama et al., 2016; Hao et al., 2019; Gagliano et al., 2016). Compounding this genetic predisposition, disruptions in the blood–brain barrier—a hallmark and risk factor of AD—allow immune cells and autoantibodies to infiltrate the central nervous system, triggering pathological immune responses such as antibody-dependent complement activation that promote amyloid-β plaque formation and microglia-induced neuronal death (Sweeney et al., 2018; D'Andrea, 2003).

In a recently suggested novel model, the AD² model reconceptualizes AD as a condition driven by dysregulated innate immunity and autoimmune-like responses rather than solely by protein aggregation (Weaver and Donald Weaver, 2023). In this framework, diverse stimuli—including infection, trauma, ischemia, pollution, metabolic syndrome, and notably, depression—act as triggers that release damage- or pathogen-associated molecular patterns (DAMPs/PAMPs). These molecular signals are sensed by innate immune receptors, initiating an inflammatory cascade. Key early events include the activation of the NLRP3 inflammasome and other pattern recognition receptors, which together foster the release of proinflammatory cytokines. Concurrently, amyloid-β, traditionally viewed as merely a pathological aggregate, is reinterpreted as an immunopeptide with antimicrobial properties. Under normal circumstances, amyloid-β serves protective roles; however, when excessively produced or misdirected by persistent immune activation, it can aggregate and further stimulate innate immune responses, perpetuating neuroinflammation.

Depression is highlighted as a possible trigger within this cascade (Weaver and Donald Weaver, 2023). Chronic depressive states are associated with prolonged stress and systemic inflammation (Kim et al., 2022), which may prime the innate immune system, enhance NLRP3 activation, and promote amyloid-β deposition. Together, these events create a self-perpetuating cycle where sustained immune activation leads to synaptic dysfunction, neuronal loss, and ultimately, the cognitive decline seen in Alzheimer’s disease.

In another attempt to frame the autoimmune hypothesis for AD, Arshavsky (2020) proposed that the innate immune response is directed toward specific proteins that are specific for engram neurons. In his explanation, Arshavsky hypothesized that creating memories might be similar to stem cell differentiation, in which epigenetic mechanisms are involved in coding long-term memories within the DNA. As such, these modifications are accompanied by protein synthesis that is unique to the engram neurons. Consequently, the adaptive immune system may identify these proteins as non-self and trigger an autoimmune response toward the engram neurons.

The emerging view of Alzheimer’s disease as an innate autoimmune condition adds a critical layer to our understanding of its pathophysiology, one that bridges immune dysregulation with profound alterations in brain metabolism. Chronic activation of the immune system, whether through microglial priming, complement cascade engagement, or sustained autoantibody responses, imposes a persistent metabolic burden on the brain Immune cells (Pathak et al., 2024), particularly microglia and astrocytes, shift toward pro-inflammatory phenotypes that rely on glycolytic metabolism, generating ROS and pro-inflammatory cytokines (Orihuela et al., 2016; Liddelow et al., 2017). Meanwhile, neurons exposed to this inflammatory milieu experience mitochondrial dysfunction, oxidative stress, and impaired nutrient utilization, hallmarks of metabolic failure observed in AD (Isaac, 2016; Swerdlow, 2017).

Neurons are highly susceptible to mitochondrial dysfunction due to their inherent characteristics. Axonal mitochondria are crucial for impulse conduction, and the length of the axon requires varying ATP levels along its length. This is managed by distributing mitochondria through multiple microtubules and neurofilaments. Ion channels and energy-dependent pumps, like sodium and potassium ATP pumps, are vital for axonal excitability and restoring baseline conditions, so any disruption in energy availability can significantly impair neuronal function. Neurons rely on glucose from the blood, regulated by the BBB and GLUT. Once glucose crosses the BBB, it serves not only as an energy source but also as a precursor for synthesizing fatty acids, other lipids, amino acids for neurotransmitters, and sugars for nucleotides.

As mentioned earlier, signs of disrupted glucose metabolism are seen early in the disease, and even decades before any clinical signs (Kumar et al., 2022). Reduced brain glucose can impair memory and synaptic plasticity as well as increase tau phosphorylation (Lauretti et al., 2017). Samples from the dorsolateral prefrontal cortex showed a wide profile of dysregulated metabolic features including bioenergetics, and neurotransmitters by-products, noting that the metabolic dysregulations seen were driven mostly by tau pathology rather than Aβ (Batra et al., 2023).

Recently, a study on postmortem AD brains found that impaired glucose phosphorylation and increased levels of several tricarboxylic acid (TCA) metabolites, except for citrate, were observed. The extent of the phosphorylation impairment and the rise in TCA cycle metabolite levels were negatively and positively correlated, respectively, with the clinical symptoms of AD (Kurano et al., 2024).

In neurons, OXPHOS is associated with increase in the production of reactive oxygen species and free radicals and results in oxidative damage, specifically to the mitochondrial membrane leading to increase in electron leak and to key enzymes in mitochondrial bioenergetics such as pyruvate dehydrogenase and α-ketoglutarate. The effect on such enzymes includes a reduction in their activity and the subsequent attenuation of the electron transport chain (ETC) efficiency. Neuronal progenitor cells derived from LOAD patients showed alterations in energy production between mitochondrial respiration and glycolysis, resulting from changes in the processing of bioenergetic substrates and the transfer of reducing agents. This included reduced levels of NAD/NADH, decreased glucose uptake, and lower response rates to insulin/IGF-1 signaling. Additionally, there was a reduction in insulin receptor and GLUT-1 densities, along with changes in the metabolic transcriptome (Ryu et al., 2021).

A study on rat AD model that underwent glucose administration after overnight fasting showed that glucose uptake in the entorhinal cortex and hippocampus was reduced, indicating a possible decrease in GLUTs in these areas (Joo et al., 2022). Indeed, the study on the autopsy program of the Baltimore Longitudinal Study of Aging showed an increase in brain glucose levels, increase in glycolytic flux and lower levels of GLUT-3 in AD brains, all of which correlated with the disease severity and hallmarks of AD pathology (An et al., 2018). Glucose-6-phosphate dehydrogenase (G6PD)is essential for the maintenance of the nicotinamide adenine dinucleotide phosphate (NADPH) pool, it acts as an oxidoreductase in the pentose phosphate pathway (PPP), a key metabolic pathway for glucose metabolism (Tiwari, 2017), and was found to be reduced in AD (Ulusu, 2015). Interestingly, G6PD overexpression in AD mice model attenuates the cognitive impairment, enhanced physical abilities, and reduced the stress load in the brain cortex (Correas et al., 2024). A recent study that investigated cerebral glycolytic metabolism in 5XFAD mice using molecular imaging found that the AD model exhibited age-related alterations in glucose uptake compared to wild type. These changes likely reflect the progressive metabolic dysfunction associated with AD pathology. The Analysis also revealed that the structure of metabolic covariance networks varied significantly with age and sex. In 5XFAD mice, there was a notable disruption in metabolic coupling, indicating that the coordination of metabolic activity among different brain regions deteriorates as AD progresses (Chumin et al., 2024). Moreover, a pilot metabolomics study on postmortem pre-frontal cortical tissue of AD patients found that energy metabolites derived from glucose in the glycolytic and PPP, as well as the ketone body β-hydroxybutyrate, were consistently reduced in the AD brain compared to the control brain (Patel et al., 2024).

Astrocytes play essential roles in the normal functioning of nervous tissue. They express numerous receptors that enable them to sense neuronal activity and clear neurotransmitters such as glutamate, ATP, GABA, adenosine, and endocannabinoids. This helps maintain synaptic transmission, prevent excitotoxicity, and provide neuroprotection. Additionally, astrocytes promote the formation of synapses by producing and secreting vital factors like cholesterol, glypicans, and thrombospondins (Sidoryk-Wegrzynowicz et al., 2010).

In AD, the role of astrocytes is widely acknowledged (Garwood et al., 2017; González-Reyes et al., 2017). APOE, a significant genetic risk factor for LOAD, is predominantly expressed in astrocytes in a healthy brain (Yu et al., 2014). In the presence of senile plaques, astrocytes become reactive, showing morphological hypertrophy with thicker processes and elevated levels of intermediate filament proteins like glial fibrillary acidic protein (GFAP), vimentin, nestin, and synemin (Escartin et al., 2021). In AD, there is often a severe disruption or complete loss of interlaminar astrocytes. Additionally, astrocytes may be involved in β-amyloid production by upregulating BACE1 and the APP in the diseased brain (Frost and Li, 2017).

Astrocytes, the primary regulators of energy in the brain, play a crucial role in maintaining brain homeostasis. They restore ion gradients, such as those involved in postsynaptic and action potentials, and manage the uptake and recycling of neurotransmitters, all of which require significant brain energy. Astrocytes are situated between blood vessels and neurons, placing them in an ideal position to regulate glucose flow into the brain and act as metabolic sensors.

Glutamate, the primary excitatory neurotransmitter in the brain, is critical for synaptic plasticity and neuronal health. Since glutamate cannot cross the BBB, it relies on glucose to act as a precursor for its synthesis in astrocytes. Under normal conditions and following transmission, glutamate is removed by astrocytes following sodium gradient and through glutamate transporters (GLAST (EAAT1), GLT1 (EAAT2)). Subsequently, glutamate is either transformed into glutamine or utilized as cellular fuel in cases of high concentrations.

It has been suggested that metabolic changes in astrocytes are triggering factors in the AD pathology, as they result in reduced Aβ clearance and mediate the increase in its production. Moreover, astrocyte metabolic dysfunction produces oxidative stress and neuroinflammation that are central to the disease progression in that it facilitates the hyperphosphorylation of tau and the formation of NFTs (Cai et al., 2017; Yan et al., 2013). Human iPSCs- derived astrocytes and neurons carrying familial AD mutations (PSEN1 or APP) showed higher levels of glycolysis and oxidative glucose metabolism along with elevated levels of glutamate synthesis. Moreover, astrocytes showed decreased expression of EAAT2, which reduced glutamate reuptake (Salcedo et al., 2024). The increase in glutamate, though may offer an early oxidative protection through maintaining glutathione content (an important antioxidant in the brain), however activated astrocytes surrounding Aβ plaques are correlated with disease progression, having a direct effect on synaptic plasticity and neurotransmission (Verkhratsky et al., 2010).

In 5XFAD mice, astrocytes showed reduced TCA cycle activity and reduced glutamine synthesis early in the disease, this attenuated neuronal GABA synthesis in the hippocampus (Andersen et al., 2021). Single-cell transcriptional profiling of astrocytes in prefrontal cortex of AD patients revealed a metabolic reprogramming characterized by reduced glycolysis and TCA, and reduced glutamate metabolism, interestingly, glutamine synthetase (GLUL) and glutamate dehydrogenase (GLUD1) expression were downregulated, these enzymes regulate key glutamate metabolism fluxes (Fan et al., 2023) and were shown to be an indicator of memory decline in AD patients (Neuner et al., 2017).

Magnetic resonance spectroscopy in AD patients showed increased levels of Myo-inositol and lactate levels, indicating a relation between activated astrocytes and altered energy metabolism in the brain, these metabolites were positively associated with clinical signs of cognitive dysfunction (Hirata et al., 2024). Interestingly, My-inositol was reported as a marker for MCI in mice, the rise in its levels was accompanied by increase in GFAP, an intermediate filament protein that makes up the framework of astrocytes cytoskeleton, and a maker of astrocytic activity (Ebert et al., 2021). Similarly in AD patients, plasma levels of GFAP correlated with Myo-inositol levels only in carriers of the APOE4 allele (Spotorno et al., 2022). Recently, it was found that the earliest increase in CSF Aβ_42/40 triggered an increase in plasma GFAP, which partially explained the association between CSF Aβ and brain Aβ accumulation measured by PET (Pelkmans et al., 2024). These findings point toward a key involvement of astrocytic activity, changes in metabolic profile and the early imbalance between soluble, and aggregated Aβ.

Microglia are the innate immune cells of the CNS, they comprise 10–15% of glial cells in the brain. Microglia can be found in either activated (pro-inflammatory, producing cytokines such as IL-1 β INF- γ, IL-8, TNF-α and IL-6) or dormient state (anti-inflammatory, producing IL-10 and IL-4). Besides their defense against pathogens, microglial processes enable them to survey the microenvironment in the brain, communications between microglia and neurons enables microglia to survey and maintain efficient neural activity by pruning old or defective neural connections (Wolf et al., 2017).

The high energy consumption of microglia is generally met through glucose metabolism. Microglia have the ability to switch between glycolysis and OXPHOS based on energy demand. In resting state, microglia produce ATP through OXPHOS, however under inflammation or stress conditions, microglia switch to anaerobic glycolysis for faster energy production (Aldana, 2019). To compensate for the energy deficiency, microglia enhance glucose absorption through GLUT-1 receptors. Studies have shown that inhibition of OXPHOS or ETC in cultured microglia/macrophages activates microglia and causes morphological changes and activation of molecular pathways such as mitogen-activated protein kinase (MAPK)/NF-κB pathway and NOD-like receptor protein 3 (NLRP3) inflammasomes, Increased production of pro-inflammatory cytokines and the rapid buildup of damaged mitochondria result in the dysfunction and apoptosis of microglia and macrophages (Li et al., 2022).

Reduced nutrient availability, including glucose, and altered mitochondrial ETC activity in Alzheimer’s disease impair energy supply, suggesting oxidative stress damage and metabolic reprogramming in microglia. It was also reported that cultured microglia chronically exposed to Aβ lead to metabolic reprogramming and entering a state of tolerance which can be reversed by INF-γ (Baik et al., 2019). This reprogramming was dependent on the mTOR-HIF-1α pathway. Interestingly, microglia isolated from 5XFAD mice and treated with Aβ showed differentially enriched biological processes related to metabolic process and pathways including carbohydrate derivative metabolic process, primary metabolic process, and carbon metabolism, indicating a clear metabolic dysregulation (Baik et al., 2019).

Moreover, activated microglia shows an increase in succinate levels caused by a break in the Krebs cycle at succinate dehydrogenase (SDH) (Mills et al., 2016). Succinate is crucial in oxidative metabolism. As an intermediate in the TCA cycle, it directly interacts with the ETC, providing an alternative way for ATP production through oxidative metabolism (Giorgi-Coll et al., 2017). Succinate is a known proinflammatory metabolite that increases during macrophages activation. LPS-activated microglia showed high glycolysis and mitochondrial respiration levels, the metabolic rewiring was reversed by inhibiting SDH, causing the cultured microglia to shift back to M2 phenotype (Sangineto et al., 2023).

Alterations in the metabolic profiles of neurons, astrocytes, and microglia (Figure 1) play a critical role in the pathophysiology of AD. These cellular changes disrupt normal brain homeostasis and contribute to neurodegeneration, inflammation, and cognitive decline observed in this disorder.