TREM2 regulates innate immunity in Alzheimer’s disease

Studies have shown that a few rare variants of TREM2 are considered to be associated with susceptibility to AD [22, 23]. Research on TREM2 missense mutations in European populations revealed that variants, such as L211P [24], H157Y [25], R136Q [26], T96K [26], D87N [27], T66M [26], R62H [28], R47H [27], and Q33X [29] (Fig. 2), have been found to be associated with AD. However, studies of non-European populations have shown different results. In our research on the Chinese population, the R47H missense variant was very rare, and another missense variant, G115S, was found to be related to AD [30].

Many studies reported that the R47H variant of TREM2 is associated with the risk of AD [9, 27, 31‐33]. Lill and her colleagues reported that the rs75932628 variant of TREM2 significantly increased the level of CSF-total-tau but not Aβ42 in a European population and suggested that the role of TREM2 in AD may involve tau dysfunction [34]. However, as shown in previous studies, the rs75932628 variant of TREM2 was not detected in either Chinese or Korean populations [30, 35‐37]. These results suggest that TREM2 is differentially associated with the incidence of AD in varying ethnicities, which may be related to the genetic backgrounds of different races.

The R47H variant of TREM2 increases terminal glycosylation of complex oligosaccharides in the Golgi apparatus and reduces TREM2’s solubility. This may affect the binding of DAP12 to TREM2, which would, in turn, affect the function of the receptor [38]. Meanwhile, the R47H variant has been presumed to destroy the stability of the TREM2 protein [39]. On the basis of crystalline structural analysis, another explanation suggested that AD risk variant R47H might impact binding to a cell-surface ligand (TREM2-L) and slightly impact the stability and structure of TREM2 [40]. However, a contrary result showed that transfected R47H-TREM2 constructs have an increased half-life relative to wild-type TREM2 and can resist proteasome degradation in the endoplasmic reticulum (ER) [38]. These statements are possible, which depend on the specific activity of life. The tyrosine-38 and threonine-66 residues of TREM2 are essential for the glycosylation of the protein. The Y38C and T66M variants of TREM2 may cause some significant differences in the glycosylation patterns and damage during transport to the plasma membrane. In the previous studies, it was found that the TREM2 R47H variant had a slight difference in N-glycosylation of the complex oligosaccharide compared to the Y38C and T66M variants, which are associated with Nasu-Hakola disease (NHD) [38]. This difference causes NHD to be an early-onset disease and AD to be a late-onset disease.

Data have shown that TREM2 plays a vital role in the cognitive function of the brain. An important function of TREM2 is its regulation of phagocytosis in microglia. Microglial removal of damaged cells, organic matrix molecules, and biomacromolecules must be assisted by the TREM2-DAP12 receptor complex. As a glial cell immunoreceptor, TREM2 has been found to modulate microglia-mediated inflammatory responses [41]. A decade ago, Gordon described the mechanism of two opposite types of macrophage activation [42], but now, the type M1 and M2 are widely used to define classically (proinflammatory) and alternatively activated (anti-inflammatory) microglia, which is controversial [43]. Outside the CNS, the mononuclear phagocyte system has been divided into M1 phenotype and M2 phenotype. Study shows that in the CNS, because that microglial activation is heterogeneous, the microglia can also be categorized into two opposite types: M1 phenotype and M2 phenotype [44]. Therefore, we conjecture that in the brain, microglia have two opposite roles, proinflammatory (M1, cytotoxic) and anti-inflammatory (M2, neuroprotective). TREM2 inhibits neurotransmitters by blocking M2 microglia. This may reveal the potential mechanism by which TREM2 inhibits microglial inflammatory responses [45]. The microglial cells participate in the removal of Aβ aggregates through phagocytosis. Previous study found that TREM2 overexpression by intracerebral lentiviral particle injection significantly reduced soluble and insoluble Aβ42 aggregates in the brain. In middle-aged APPswe/PS1ΔE9 mice (7–8 months old), the ability of microglia to remove amyloid plaques increased after TREM2 overexpression, and the density of amyloid plaques in the brain decreased [20]. However, in a mouse model of TREM2 defects, the concentration of amyloid plaques in the brain did not change [46]. Interestingly, after the expression of TREM2 in 18-month-old APPswe/PS1ΔE9 mice, the concentration of amyloid plaques was not attenuated, and no alterations in the levels of Aβ42 were observed in the brain [46]. Research utilizing mouse models has shown that the overexpression of TREM2 plays a protective role in both early- and mid-term AD, whereas this protective effect is lost in late-term AD [47]. We speculate that the reduced number of microglia in the brains of older mice may lead to a decline in phagocytosis.

The high level of phosphorylation and abnormal aggregation of tau protein are pathophysiological factors associated with neuronal and synaptic damage. The loss of neurons and synapses in the hippocampus is associated with a decrease in spatial cognitive function. The 7-month-old P301S mouse model has been shown to exhibit significant neuronal and synaptic damage to this region. Overexpression of TREM2 is effective in inhibiting these lesions; water maze experiments have demonstrated that TREM2 overexpression can restore spatial cognitive impairment in mice [47]. In addition, the overexpression of TREM2 by intracerebral lentiviral particles injection has been found to significantly improve hyperphosphorylation of tau proteins and reduce the activity of cyclin-dependent kinase 5 (CDK5) and glycogen synthase kinase-3β (GSK3β) [45]. Thus, TREM2 overexpression significantly reduces neuronal loss and may play a role in the phosphorylation of tau protein, thereby reducing the incidence of AD.

A recent study showed that TREM2 releases its extracellular domain after protease cleavage, leaving only the carboxy-terminal fragment (CTF) attached to the membrane [48]. Soluble TREM2 (sTREM2) may be produced by proteolytic cleavage and alternative splicing. If insertions [49] or frameshifts [50] occur in exon 4, it can terminate the transmembrane domain, which is speculated to yield a soluble product. In addition to the membrane-bound form, sTREM2 has been detected in the supernatants of human [51] and mouse [51] cell cultures and in the peripheral blood and cerebrospinal fluid (CSF) [52]. The sTREM2 in human peripheral blood and CSF can be used as a more accurate tool for understanding the biological effects of TREM2 in the pathogenesis of AD. Hu et al. analyzed the expression of TREM2 mRNA and protein in the peripheral blood in a population of Northern Han Chinese [53]. The results showed that on the level of mRNA and protein, TREM2 expression were higher on monocytes, granulocytes, and in plasma in AD group compared with that of control groups. Mori et al. performed a similar analysis of TREM2 expression in the peripheral blood in a small population of Japanese individuals (26 patients with AD, 8 males and 18 females) [54]. However, another study mentioned that the absolute level of TREM2 expression in human peripheral blood monocytes is quite low and unlikely to be useful for drawing mechanistic conclusions about TREM2 [55]. The upregulation of TREM2 in the peripheral blood indicates that the gene is abnormally active in the development of AD pathology. More experiments are needed to confirm whether TREM2 was differently expressed in the peripheral blood in some populations. The level of sTREM2 in the CSF also exhibits changes. Although Kleinberge et al. showed that sTREM2 levels were reduced in the CSF of AD patients [56], other studies have shown that sTREM2 levels in the CSF increased with age and were positively correlated with the levels of Aβ42 and tau protein [26, 57‐59].

While most people think that TREM2 exerts an anti-inflammatory effect, it seems that the connection between TREM2 and other inflammatory responses is not so simple. According to the cell type and context, the strength [60] and duration [61] of the stimuli is different. Therefore, TREM2 seems to play different roles in inflammatory responses.

Some in vitro and in vivo studies have shown that TREM2 plays an anti-inflammatory role in certain contexts. In cell lines, TREM2 deficiency increases the levels of proinflammatory mediators, such as tumor necrosis factor-α (TNFα), interleukin-1β (IL1β), and interleukin-6 (IL6) [62]. TREM2 knockdown in the senescence-accelerated mouse P8 (SAMP8) mouse model also increased the production of inflammatory cytokines [63]. Furthermore, overexpressing TREM2 in AD mouse models [20, 45] reduced the levels of proinflammatory transcripts. From these studies, we can speculate that TREM2 can inhibit inflammatory responses in some contexts.

However, many studies have supported that TREM2 can amplify or promote inflammatory responses. TREM2-deficient microglia have reduced activation and a more ramified morphology in cell cultures [19]. In AD mouse models, TREM2-deficient microglia exhibit decreased cell size and surface area, as well as increased process length, resulting in reduced activation [16]. sTREM2 activate the Akt–GSK3β–β-catenin pathway, which can suppress apoptosis in microglia [64]. In this study [65], TREM2 promotes microglial survival by activating the Wnt/β-catenin signaling pathway. The upregulation of the Wnt/β-catenin pathway suppresses GSK3β, restores β-catenin signaling, and promotes TREM2-deficient microglial survival in vitro and in vivo. NF-κB signaling is associated with proinflammatory cytokines; the inhibition of NF-κB signaling markedly downregulated the production of three proinflammatory cytokines (IL-1β, IL-6, and TNF) [64]. Taken together, these findings clearly supported that TREM2 can regulate inflammatory responses.

Aβ can destroy synaptic transmission, induce oxidative stress, and trigger cell death in vitro [66]. Meanwhile, microglia have been shown to devour Aβ in the brain [67]. Therefore, microglial phagocytosis of Aβ may serve a neuroprotective function. However, the absence of TREM2 significantly impairs the ability of microglia to engulf amyloid plaques. Some studies have reported that a TREM2-deficient AD mouse model results in a decrease in the number of microglia around amyloid plaques because metabolic fitness is reduced [68].

Condello et al. proposed a new hypothesis [69]; they postulated that the tight envelope of microglia around the amyloid surface constitutes a neuroprotective barrier that limits fibril outgrowth and plaque-associated toxicity. In AD mouse models, a lack of TREM2 or DAP12 results in more dispersed amyloid plaques and increased synapses, causing a morphology that resembles a sea urchin [17]. The greater the number of synapses that protrude outside is, the larger the contacted surface with nerve structures and the greater the potential harm to the nervous system. Thus, in human brains, the protective role of microglia may primarily act as a barrier that isolates amyloid plaques from peripheral nerve tissues.