A$\beta$ is found in various assembly forms, such as fibrils, protofibrils, and oligomers [19]. A$\beta$ is non-neurotoxic in its monomeric form. In contrast, protofibrillars and oligomers are regarded as effective blockers of long-term potentiation and strength. Fiber formation is strongly linked to protein misfolding [20].

A$\beta$ induces fatality in neuronal cells, which is the primary cause of AD [21]. Significant evidence has shown that the production of A$\beta$ oligomers (AbOs) in cortical neurons initiates AD [22, 23, 24]. AbOs acting on entorhinal neurons assist in tau oligomer development with progression to associated neurons. This occurs in the hippocampus and subsequently in the subiculum and associated neocortex [25]. The toxicity of tau oligomers can be damaging to neurons, where they are formed and subsequently released into neurons and synapses. Tau proteins are propagated and disseminated from neurons. AbOs may interact with tau oligomers during this phase to enhance neuronal and synaptic dysfunction [26]. Data further demonstrates that A$\beta$ is not a bystander in AD [27]. Recent advances have revealed that extracellular A$\beta$ deposits can be released from the brain via the blood-brain barrier (BBB) and other systems [28]. The structure and function of the BBB are damaged in AD [29]. The destruction of the BBB allows neurotoxic blood-derived debris and microbes to enter the brain, which is closely linked to neuroinflammation and immune responses that can trigger numerous neurodegenerative pathways [30, 31].

Tau pathology is a hallmark of AD progression [32]. Tau present in the medial temporal lobe binds and stabilizes the microtubule architecture, where it is normally concentrated. However, hyperphosphorylated tau protein produced in AD dissociates from microtubules and accumulates inside neurons [33]. Once it spreads from the medial temporal lobe to the surrounding neocortex, hyperphosphorylated tau protein can cause cognitive impairment.

Findings from animal models have shown that A$\beta$ spreads via neuronal connections from cell to cell [34, 35]. An increasing body of evidence demonstrates that the synergistic effects of A$\beta$ on tau plays a vital role in AD pathogenesis [36, 37]. Busche et al. [38] recently showed that A$\beta$ and tau interaction promotes neural circuit damage: in vivo multiphoton imaging showed that the combined existence of tau and A$\beta$ pathology in the neocortex is linked to repressed neuronal activity and increase in spine loss and microglia injury. The repression of tau or A$\beta$ pathology alone is not conducive to salvaging functional damage. In addition, the impact of tau and A$\beta$ co-expression on neuronal activity was analyzed in Busche and Angulo’s studies. Their findings [39] showed that extracellular expression of A$\beta$ caused hyper-excitability, whereas tau expression led to activity suppression. The co-expression of A$\beta$ and tau also repressed activity, where the tau phenotype appeared to play a dominant role.

Cholinergic neurons have been demonstrated to modulate neuronal circuits in the hippocampus and cortical regions and play an important role in hippocampus-dependent memory [40, 41]. Normally, endogenous nerve growth factor (NGF) is produced by hippocampal and post-synaptic cortical neurons and is expressed through related receptors present on presynaptic cholinergic terminals [42, 43]. Stimulated NGF is then transported from the target regions (hippocampus and cortex) to cholinergic nuclei, consequently initiating cholinergic signaling in these brain areas [44].

Findings from various animal models and human studies have shown that cholinergic neuron degeneration in the basal forebrain is linked to cognitive impairments in AD [45, 46]. The cholinergic hypothesis primarily focuses on the gradual damage of limbic and neocortical cholinergic innervations, which is linked to neurofibrillary degeneration and inefficient axonal transport and signaling [47]. The crosstalk between postsynaptic cortical/hippocampal neurons and presynaptic cholinergic terminals is altered during AD. These variations are related to the accessibility of mature NGF (mNGF) to basal forebrain cholinergic neurons, which subsequently results in cholinergic degeneration in hippocampal and cortical areas [48, 49].

The cholinergic hypothesis has transformed different areas of AD research, from the field of neuropathology to the modern concept of synaptic neurotransmission [50]. It was developed based on three parameters: (1) diminished presynaptic cholinergic indicators in the cerebral cortex [51, 52], (2) the nucleus basalis of Meynert (NBM) in the basal forebrain is a source of cortical cholinergic innervation that undergoes intense neurodegeneration in AD [53, 54] (3) cholinergic antagonists impair memory whereas agonists have a contrasting effect on memory [55].

GABAergic activity is important for brain development and plasticity. Under normal conditions, GABA is produced in the presynaptic membrane and helps synthesize glutamic acid decarboxylase (GAD) from glutamate. GABA can bind to GABA receptors present on the postsynaptic membrane through the synaptic cleft, which inhibits the post-synaptic neuron [56, 57]. However, in AD, imbalances in GABAergic neurons is associated with several disorders. In the AD brain, A$\beta$ is toxic to GABAergic synapses and promotes the degeneration of axons and synaptic loss. At first, A$\beta$ enhances neural activity, which supports A$\beta$ release. The cycle between neural activity and A$\beta$ can involve A$\beta$-induced GABAergic terminal loss, which results in impaired cognition and repressed long-term potentiation (LTP). In addition, $\beta$-site APP cleaving enzyme 1 (BACE1), Apolipoprotein E4 (APOE$ϵ$4), and TREM2 can induce GABAergic dysfunction [58].

These outcomes have shown that GABAergic dysfunction is related to the E (excitation)/I (inhibition) imbalance in AD [59]. As mentioned Govindpani et al. [60], because of the narrow scope of existing treatments for disease modification, better insights into GABAergic remodeling in AD may help introduce innovative and unique therapeutic options.

In recent decades, AD investigations have prioritized the mechanisms related to extracellular neural plaques (NPs) and intracellular neurofibrillary tangles (NFTs). Data from genetic, preclinical, and clinical findings have focused on the neuroinflammatory mediation of the innate immune system [66]. Many neurodegenerative diseases, particularly AD, have been linked to inflammation [67, 68]. Increasing evidence suggests that systemic inflammation plays a key role in AD [69]. Tau-protein tangles and A$\beta$ plaques in the brain trigger cells, such as microglia and astrocytes, to release anti-inflammatory and pro-inflammatory mediators, including reactive oxygen species (ROS), neurotransmitters, chemokines, and cytokines [70, 71]. The release of mediators causes lymphocytes and monocytes to pass through the BBB, which boosts the release of inflammatory components.These findings indicate that inflammation promotes AD progression and accelerates disease progression [72, 73]. Trovato et al. [74] eported that an increase in oxidative stress and altered antioxidant systems cause NOD-like receptor thermal protein domain associated protein 3 (NLRP3) inflammasome activation, resulting in cell damage and cognitive impairment.

Microglia are the primary macrophages and provide basic immunological protection in disorders and injuries of the central nervous system [75, 76]. Plastic cells have dual functions in neuronal damage and healing and can adopt multiple phenotypes [77, 78]. Microglia have a ramified phenotype under normal physiological conditions, characterized by a small cell body and several processes, such as in “resting microglia”. Even in the resting state, microglial processes are dynamic and constantly scan to preserve central nervous system (CNS) cell types with neurons [79]. Evidently, they scan the brain environment continuously and contact synapses by using their fine branches to detect infections and damage in their environment [80].

Microglia, which represent the immune cells of the brain, engage in diverse functional programming, known as polarization, for responding to external stimulating factors [81]. In the activated state, microglial phenotypes are categorized by two major states: classical activation (pro-inflammatory microglia) and alternative activation (anti-inflammatory microglia) [82, 83]. Pro-inflammatory microglia primarily release proinflammatory substances, whereas anti-inflammatory microglia release anti-inflammatory cytokines. When triggered by lipopolysaccharide (LPS), the pro-inflammatory phenotype is acquired by microglia, which leads to neurotoxicity via secreted pro-inflammatory factors such as tumor necrosis factor-$\alpha$ (TNF-$\alpha$) and Interleukin 1 (IL-1) [83]. In contrast, microglia acquire an anti-inflammatory state via the release of anti-inflammatory cytokines (e.g., Interleukin 4.), which aids neuroprotection and promotes tissue repair, both of which play crucial roles in maintaining the physiological environment [84, 85].

In neurodegenerative conditions, especially in AD models, persistently activated microglia may restrict CD3${}^{+}$/CD8${}^{+}$ T-cell entry into the brain. Local macrophages constitute a link between innate and adaptive immunity [86]. In response to A$\beta$ aggregation, including disease progression, the production of proinflammatory cytokines downregulates the expression of A$\beta$ clearance components and promotes A$\beta$-mediated neurodegeneration and A$\beta$ aggregation [87]. The early activation of microglia during AD progression provides neuroprotection by supporting A$\beta$ clearance. However, pro-inflammatory cytokines promote A$\beta$ aggregation during disease progression. Evidence from multiple animal studies provides strong support for the production of A$\beta$, which increases AD deficits by upregulating microglial activation. A$\beta$-excess neurons, as primary proinflammatory factors, implicate the intraneuronal aggregation of A$\beta$ as an important immunological element in AD pathogenesis [88].

Microglia show different phenotypic states in AD modesl, especially under chronic inflammatory conditions [89]. Some receptors (receptor for advanced glycation endproducts [RAGE] and NLRP3) of the proinflammatory microglial phenotype secrete proinflammatory cytokines, such as TNF-$\alpha$ and IL-1$\beta$, by triggering a signaling cascade to induce neuronal cell death. Meanwhile, other receptors (the class a macrophage scavenger receptor type I [SR-AI], TREM2) of the M2-like microglia participate in clearing A$\beta$ by stimulating A$\beta$ fibril internalization and synthesizing anti-inflammatory cytokines such as TGF-$\beta$, IL-10, and IL-4 [90, 91]. Activated microglia play a key role in enhancing the spreading of tau protein in the presence of A$\beta$. A$\beta$ in its soluble form, along with additional elements, can activate microglia via microglial surface receptors. Tau is taken up by activated microglia and released into further bioactive types. The released tau can be taken up by neurons and subsequently released into the neuropil, in an activation-dependent manner [92]. Therefore, finding a method to transform pro-inflammatory microglia into anti-inflammatory microglia has a prospective advantage in treating AD.

TREM2 is an innate immunological receptor found only on the surface of myeloid cells in the brain, such as microglia, macrophages, and monocytes [95, 96]. Mature TREM2 protein weights approximately 40 kDa. In situ hybridization and immunohistochemical labeling have been used to detect the protein [97]. TREM2 has a long ectodomain that interacts with the extracellular setting to control microglial function [98].

Current research indicates that minor changes in soluble TREM2 (sTREM2) levels (approximately 7%–10%), are sufficient for modulating AD risk [99]. According to Piccio et al. [100, 101], AD and other inflammatory central nervous system diseases have considerably higher CSF sTREM2 levels than normal controls. Deming et al. [102] discovered for the first time that, as a vital contributor in TREM2’s biological processes, the membrane-spanning 4-domains subfamily A (MS4A) gene cluster plays a significant role in regulating soluble TREM2 expression as well as AD pathology.

TREM2 is crucial for cell maturation, proliferation, and survival during development under homeostatic conditions [103, 104]. Multiple TREM2 functions have been identified in the last decade, including the regulation of phagocytosis and modulation of inflammatory signaling [105, 106].

Genetic variations in TREM2 have been linked to several neurodegenerative diseases [121, 122]. Recent research has shown that microglial TREM2 expression is associated with AD pathology. The most common AD-associated TREM2 variant is rs75932628, a single-nucleotide polymorphism encoding an arginine-to-histidine missense substitution at amino acid 47 (R47H), which significantly increased sporadic AD incidence and impaired phospholipid ligand binding [123, 124]. Recent studies have proposed that R62H and R47H are partial loss-of-function variants and predominantly lead to minimized affinity for TREM2 ligands [125]. In contrast, upregulated TREM2 signaling can lead to several possibilities for immune-linked AD treatments [126, 127]. The regulation of TREM2 in AD represent a novel therapeutic approach that is currently under investigation in clinical trials [128, 129]. Therefore, microglial TREM2 is a potential therapeutic target for ameliorating neurodegeneration disease. However, caution should be exercised when targeting TREM2 as a therapeutic entry point for AD until its involvement in tau aggregation and propagation is better understood [130].

There is plenty of literature demonstrating that regular physical exercise may slow disease progression or ameliorate symptoms in AD by increasing cerebrospinal biomarkers [163] and cardiovascular fitness [164], as well as decreasing AD-related biomarkers [165]. In addition, various animal models have demonstrated that exercise exerts neuroprotective effects on cognition in AD, such as object recognition memory [166], spatial learning and memory [167], and anxiety behavior [168]. Dao AT et al. [169] demonstrated that AD-impaired basal synaptic transmission and suppression of early-phase long-term potentiation in the dentate gyrus was prevented by prior moderate treadmill exercise. Using a transgenic model, Mu L et al. [170] demonstrated that strengthening structural synaptic plasticity may represent a potential mechanism by which treadmill exercise prevents decline in spatial learning and memory and synapse loss in 3 $\times$ Tg-AD mice. Liu Y et al. [171] reported that short-term resistance exercise inhibits neuroinflammation and attenuates neuropathological changes in AD mice. Lu Y et al. [172] demonstrated that treadmill exercise is neuroprotective and regulates microglial polarization and oxidative stress in AD model. Furthermore, pharmacological mimetics of exercise, such as by enhancing adult hippocampal neurogenesis (AHN) and elevating brain derived neurotrophic factor (BDNF) levels, may improve cognition in AD. Furthermore, when applied at early stages of AD, these mimetics may protect against subsequent neuronal cell death [173].

The interaction between microglia and neurons, mediated by TREM2, has been shown to contribute to AD pathogenesis. Therefore, the regulation of TREM2 to alleviate AD has garnered major interest in AD research. Recently, findings from an animal study showed that upregulation of TREM2 ameliorated inflammatory responses, neuronal injury, and cognitive deficits [174, 175]. More importantly, exercise can be viewed as a safe and economic option for improving cognitive performance in both normal and diseased states, including AD. However, few studies have revealed the critical role of TREM2 in exercise and AD.

Currently, the mechanism underlying the exercise-induced neuroprotective effect on cognitive function associated with TREM2 in an AD model can be illustrated from three major findings. First, exercise has been shown to regulate microglial function through TREM2 regulation in the brain. Improvement in recognition memory in an AD rat model is linked to the upregulation of the hippocampal TREM2/DAP12 pathway, which can inhibit excessive microglial activation as well as neuroinflammation [176]. Recent research using the APP/PS1 mouse model demonstrated that long-term running inhibits TREM2 shedding to maintain the levels of TREM2 protein as well as microglial metabolic activity [177]. Second, exercise can increase dendritic spine density by modulating TREM2 expression. TREM2 deficiency was shown to exacerbate dendritic spine loss in an AD mouse model [162]. In contrast, Mu et al. [170] showed that exercise protects memory function by enhancing dendritic spine density in the brain in a 3 $\times$ Tg-AD mouse model. Thus, it was widely speculated that exercise modulates TREM2 expression to protect dendritic spines in AD models. However, the in vivo/in vitro evidence should be validated further to confirm this hypothesis. Third, the enhancing effect of physical exercise onsTREM2 was measured in the cerebrospinal fluid of patients with AD. The experiments performed by Jensen et al. [178] indicated that the levels of sTREM2 in the cerebrospinal fluid of patients with AD increased with physical exercise intervention.