AD is a neurodegenerative disease characterized by gradual deterioration of cognitive ability, with notably gradual death of neurons and often severe impairment of cognitive ability in memory and learning. It is well known that AD is a prolonged neurological disease with its development divided into three stages, i.e., preclinical stage, mild cognitive impairment (MCI) stage, and demented stage (7, 8). In about 20 years ago, the “amyloid cascade hypothesis” was proposed to explain the pathogenesis of AD based on the formation of abnormal Aβ plaques in different regions of the brain, suggesting that the abnormal accumulation of Aβ outside the cells was the main cause of the cascade effect leading to neuronal injury (9, 10). However, the accuracy of Aβ as a biomarker for AD is seriously challenged because Aβ is found in approximately 30% of cognitively normal elderly people (11). Similarly, the results of Aβ clearance as a treatment for AD patients are not satisfactory, indicating that the “amyloid cascade hypothesis” is unable to fully explain the molecular mechanism of AD pathogenesis (12, 13). Furthermore, another significant pathological change in AD is the intracellular aggregation of neuronal fibrillary tangles formed by tau aggregation (14). Moreover, several alternative hypotheses, i.e., the oxidative stress hypothesis, the cholinergic hypothesis, and the neuroinflammation hypothesis, are also proposed to provide significant insights into the pathogenesis of AD (15–17). Among these hypotheses, the neuroinflammation has been shown to be involved in the entire process of AD. Growing evidence suggests that persistent glial cell-mediated neuroinflammation is the primary cause for both neurodegenerative processes and cognitive deficits in AD patients. Inflammation is not only a passive outcome but also a contributor to the occurrence of AD, attracting increasing attention from medical scholars to further explore the important functions of neuroinflammation in the development of AD (18).

Neuroinflammation is a reaction of inflammation in the CNS due to pathological damage occurring in the peripheral or CNS, resulting in the production of proinflammatory cytokines, such as interleukin-1β (IL-1β), IL-6, IL-8, and tumor necrosis factor (TNF), chemokines, complement cytokines as well as some small molecule messengers, e.g., prostaglandins, nitric oxide (NO), and reactive oxygen species (ROS) (19). Generally, the acute inflammation in the brain exists as a defense against damage to the nervous system, infection, and other stimuli. However, with prolonged inflammation, the acute inflammation gradually becomes chronic to stimulate the nervous system. The chronic inflammation could continue to release various inflammatory cytokines, resulting in a proinflammatory response that is consistently stronger than the anti-inflammatory response, thus aggravating the inflammation, ultimately damaging the neurons and causing various pathological changes in the body (20). Endogenous bioactive lipids regulate a large number of cellular and molecular processes related to health and disease, especially during inflammation. All major endogenous bioactive lipids, i.e., eicosanoids, specialized pro-catabolic lipid mediators, lysophospholipids, and endocannabinoids, are closely associated with chronic inflammation (21). Although these molecules are originally described as neuromodulators/neurotransmitters, they have been characterized as important regulators of glial function (22) and will in turn disrupt tissue homeostasis to cause chronic damage, such as AD (23). Given the higher expression and production of inflammatory cytokines in microglia than those of other glial cells, microglia are considered the main sources of brain inflammation. Furthermore, the microglia have attracted increasing attention from the medical scholars due to their complex phenotypes. Moreover, microglia are also closely related to the loss of synapses, tau phosphorylation, and poor memory, making these biological processes increasingly the centers of the study of neuroinflammation in AD (24).

In the rapidly changing environment of brain, precise differentiation of microglial phenotypes is essential to explore the pathological development of AD. The phenotypic characters of microglia could be divided into resting phase (M0) and activated phases (M1 and M2) (31, 32) ( Figure 1 ). In the resting microglia, cytosol remains in a fixed position. The two-photon imaging of the cerebral cortex reveals that microglia at the resting phase are still functioning as a sensitive surveillance, with their cytosols and protrusions continuously monitoring changes in the microenvironment of the brain parenchyma (33). The cytosols of microglia at this phase are mostly highly branched, i.e., these cells are known as multipolar or branched microglia, which are particularly sensitive to the homeostasis disturbance of the internal brain environment. In AD, the brain homeostasis is impaired, accompanied by changes in both brain microenvironment and glial cell characteristics, while the microglia mainly show the activated anti-inflammatory phenotype M2, featured by the release of anti-inflammatory cytokines, including transforming growth factor-β (TGF-β), IL-4, IL-10, and IL-13, and enhanced phagocytosis (34). The persistent inflammation transforms microglia to the activated proinflammatory phenotype M1, which promotes inflammation and leads to increased concentrations of proinflammatory cytokines, including TNF-α, IL-4, IL-6, IL-12, and IL-18, accompanied by impaired phagocytosis (35), while the microglial cell body is enlarged, the processes become shorter, and the outline of microglial cells is changed to round (33). These polarized states differ in triggering stimuli, phenotypic markers, and expression of secretory mediators ( Figure 1 ). In AD, the neuroinflammation is increased with the disease progression. Specifically, the neuroinflammation level reaches the first peak in the early development of AD, attributed to the initial anti-inflammatory response, and then a second peak during the transition from stage MCI to stage dementia, suggesting the change of microglia to a proinflammatory phenotype (36, 37). A longitudinal study revealed that microglial activation may attempt to repair damage during the initial stages of AD, while in LOAD, microglia could be harmful, producing proinflammatory molecules to cause neuronal damage (36). For example, microglia initially phagocytose apoptotic cells through Anexelekto (Axl) and Mer anti-inflammatory receptors (38), whereas the chronic microglial activation could lead to severe pathological changes and neurological complications. It is worth noting that although the M1/M2 nomenclature is rather useful in experimental paradigms, it has also been fiercely discussed, especially in microglia: in particular, M1- and M2-like phenotypes can be considered two extremes of a rather blurred distribution, rather than 2 distinct populations (39). Some authors strongly avoided this dichotomy, whereas others acknowledged that they still use it even with its limitations for lack of a better term (40). Dark microglial (DM) is a recently described phenotype, also known as “dystrophic” microglia, which was shown to precede the spread of Tau pathology (41, 42). Its dark appearance is detected under the high spatial resolution transmission electron microscopy (43). In the APP/PS1 mouse model, black microglia with complete and clear abnormal chromatin and dark appearance are observed. Compared with C57BL/6 mice, the density of black microglia in APP/PS1 mice is significantly increased and concentrated near both Aβ plaques and dystrophic neurites (44). The molecular mechanisms underlying the functions of different types of microglia remain to be explored.

Studies have shown that the microglial phenotype regulation depends primarily on the relationship between pattern recognition receptors (PRRs) and molecules released from the surrounding cells (60). In AD, these PRRs could sense Aβ, pathogen-associated molecular patterns (PAMPs), or other damage-associated molecular patterns (DAMPs) to influence the microglial phenotype. Microglia activation is characterized by enhanced responsiveness to various inflammatory factors or injury stimuli. After the microglia perceive the DAMPs and PAMPs through toll-like receptors (TLRs), retinoic acid-inducible gene-I (RIG-I)-like receptors (RLRs), nucleotide-binding oligomerization domain (NOD)-like receptors (NLRs), or other PRRs, the signaling through PRRs could trigger an inflammatory response and proinflammatory cytokine secretion ( Table 1 ). Generally, these activated microglia are called disease-associated microglia (DAM) (73–77). Furthermore, the Aβ binding particularly to TLR4 has been shown to activate microglia (63, 78). The main receptors expressed by activated microglia include scavenger receptors (SRs), TLRs, triggering receptors in myeloid cells 2 (TREM-2), CD14, CD47, Fcγ receptors (FcγRs), CD200 receptor (CD200R) and receptor for advanced glycation end products (RAGE). Specifically, CD14, TLR2, and TLR4 are required for fAβ-stimulated microglial activation (79), though the molecular mechanisms regulating this microglial activation remain unclear. However, this sustained activation of microglia is adverse because the extended activation of TLR2 and TLR4 in microglia could induce the secretion of Aβ (80). TREM2 signaling appears to be required for the maintenance of microglial metabolism, while the deficiency of TREM2 leads to impaired microglial survival, migration and, phagocytosis (81). In addition, the mild activation of TREM2 has shown anti-inflammatory function to inhibit TLR4 receptor-induced proinflammatory response (82–84). However, the role of TREM2 in the inflammatory process is still controversial. For example, the injection of AD-Tau into the brain of mice with amyloidosis to induce Aβ-dependent Tau deposition showed that TREM2 consistently activated microglia and aggravated Tau pathology as well as neurodystrophy (85). Furthermore, other Aβ peptides, fibrils, and APPS are also potent activators of microglia (86). In AD pathology, the PRRs bind to different types of Aβ with different affinities to initiate the activation of glial cells, causing accumulation of more microglia and then increasing the density of microglia (87). Moreover, Aβ could activate the NF-KB-dependent pathways to bind to the surface of microglia to activate extracellular signal-regulated kinase (ERK) and mitogen-activated protein kinase (MAPK) pathways, ultimately triggering the proinflammatory gene expression (88, 89). An autopsy study of the temporal neocortex in 15 control subjects without dementia and 91 AD patients has indicated that in the postmortem brain of AD patients, even after the Aβ plaque growth ends, the density of microglia is increased linearly and positively correlated with NFTs burden, which is consistently observed in the non-amnesic AD variant of primary progressive aphasia (54, 90). Inconsistent with other studies, the activation of Wnt-3a in primary microglia could promote microglia into the proinflammatory state of AD (91–93). Furthermore, the exogenous interfering factors, e.g., lipopolysaccharides (LPS), injected intraperitoneally in 5×FAD mice and APP23 mice are subsequently observed in the microglial hyperresponsiveness around the dense core plaque (94). Due to its significant biological importance, the activation of microglia caused by inflammation which is induced by overactivated microglia has attracted increasing attention to explore the molecular mechanisms regulating the microglial phenotypic variations.

Various factors induce imbalances of the CNS, e.g., aging, systemic inflammation, or stress, leading to the initiation of microglia and causing microglia to show strong susceptibility and responsiveness toward inflammatory stimuli. Therefore, a potential treatment strategy for AD could be to inhibit pre-disease microglial priming. Studies have shown that the middle-aged AD patients with obesity and insulin-resistance are prone to inducing inflammation, while statins exhibit a protective effect against this injury, preventing the initiation of microglia (111, 112). Furthermore, supplementation with folic acid and omega-3 fatty acid has also been shown to intervene the neuroinflammation and reduce the level of inflammation in both the cerebrospinal fluid and blood of individuals with AD (113, 114).

The factor of NF-κB has been implicated in physiological processes related to signaling, cognition, and memory in CNS. This factor could be activated by oxidative stress, neuroinflammation, and other factors, leading to CNS dysfunction in AD patients (115). The production of ROS initiates the enzymatic activities of IKKb to phosphorylate the heterodimer of NF-κB, which is an inhibitor of kappaB (IκB), leading to its degradation via the ubiquitin-proteasome pathway, while the dissociation of IκB from the dimer could initiate the NF-κB influx into the nucleus (116). It is indicated that the activation of NF-κB stimulates the BACE1 expression and Aβ processing, which is probably a novel molecular mechanism regulating the AD progression (117). It has been discovered that inhibition of NF-κB activation by antioxidants, polyphenols, and non-steroidal anti-inflammatory drugs (NSAIDs) is essential to reduce the load of Aβ (118). For example, Tanshinone I is used to inhibit the synthesis and expression of several proinflammatory M1 mediators, significantly suppressing the LPS-induced activation of NF-κB in microglia (119), while Forsythia B regulates the neuroinflammation in APP/PS1 mice, effectively ameliorating cognitive function, reducing the accumulation of Aβ and tau, and attenuating glial cell activation in the hippocampus (120). Similarly, when the NF-kB signaling pathway is inhibited, the release of pathogenic tau by primary microglia could be reduced to rescue the autophagy defect of microglia and reduce the neurotoxic reaction (109).

The NOD-like receptor thermal protein domain associated protein 3 (NLRP3) are highly enriched in microglia comparison to both NLRP1 and NLRP2 (121). Therefore, it is worth further exploring the potential application of NLRP3 in the diagnosis and treatment of AD compared with other types of inflammatory vesicles. Indeed, there is ample evidence to show that the Aβ level in brains of APP/PS1 or TgCRND8 mice could be reduced with the memories modified following the use of small molecular inhibitors of NLRP3 inflammatory vesicles or knockdown (KO) of NLRP3 expression (122, 123). Furthermore, inhibition of NLRP3 inflammasome reduces IL-1β secretion and limits the injury to neighboring tissues, while IL-1β is thought to enhance the expression of s100β protein, which is closely associated with AD (124, 125). For example, benzylisothiocyanate (126), stavudine (127), and artemisinin (128) all have shown therapeutic effects on AD mice by inhibiting NLRP3 activation.

Microglia support the neuronal function by removing toxic damage during the early stages of AD. Interference with the activation of microglia could allow their anti-inflammatory effect to persist for a long time. For example, the activated PPPAR-γ could inhibit the inflammatory response, while in mouse model of AD, treatment with PPAR-γ activators, e.g., pioglitazone and rosiglitazone, could switch microglia to show anti-inflammatory phenotype with phagocytosis (37, 129–131), which is probably related to the phagocytosis of amyloid deposits.