Emerging roles of innate and adaptive immunity in Alzheimer’s Disease

eTOC Blurb

Chen and Holtzman review the emerging roles of both innate and adaptive immunity, their alterations, interplay, and contributions to the development and progression of Alzheimer’s disease. They proposed that targeting both dysregulated innate and adaptive immune responses in brain parenchyma as well as border structures could serve as important therapeutics for preventing and treating the disease.

Summary

Alzheimer’s disease (AD) is the most common neurodegenerative disease with characteristic extracellular amyloid-β (Aβ) deposition and intracellular accumulation of hyperphosphorylated, aggregated tau. Several key regulators of innate immune pathways are genetic risk factors in AD. While these genetic risk factors as well as in vivo data point to key roles for microglia, emerging evidence also points to a role of the adaptive immune response in disease pathogenesis. We review the role of innate and adaptive immunity, their niches, communication, and contributions to AD development and progression. We also summarize the cellular compositions and physiological functions of immune cells in the parenchyma together with those in the brain border structures that form a dynamic disease-related immune niche. We propose that targeting both innate and adaptive immune responses in brain parenchyma and border structures could serve as important therapeutic targets for treating both pre-symptomatic as well as the symptomatic stages of AD.

Introduction

The concept of “immune-privilege” originated centuries ago from observations that the survival of ectopically transplanted foreign tissues or tumors survived without immunosuppression in the brain parenchyma but not the peripheral tissues. Interestingly, when tissue was first grafted peripherally and then grafted into the brain, it was rejected (Medawar, 1948; Murphy and Sturm, 1923). Albeit the homeostatic brain parenchyma might not be a typical location for most immune cells, after peripheral priming and alteration in the central nervous system (CNS) environment, non-resident immune cells are capable of migrating into and surviving within the brain parenchyma (Mrdjen et al., 2018). The entire CNS is organized into different compartments, the parenchyma, the subarachnoid space containing choroid plexus and cerebrospinal fluid (CSF), the meninges and brain skull bone marrow (Korin et al., 2017; Louveau et al., 2015a; Prinz et al., 2011; Ransohoff et al., 2003). The “immune-privileged” state appears to be conditional and has regional variability.

It is now well-recognized that cells of both the innate and adaptive immune system are present in the homeostatic CNS. In the homeostatic brain, microglia constantly surveil the brain parenchyma. There are low levels of both adaptive and innate immune cells such as T cells, B cells, neutrophils, monocytes and their derived macrophages, that can enter the brain and particularly the meninges and play a role in brain development and learning (Engelhardt et al., 2017; Filiano et al., 2016; Herz et al., 2021; Pasciuto et al., 2020; Radjavi et al., 2014; Ribeiro et al., 2019). Recent studies have shown that adaptive and innate immune cells can populate the dura via skull channels that connect the skull bone marrow to the dura (Brioschi et al., 2021; Cugurra et al., 2021). Therefore, under homeostatic conditions, immune cells can access particularly the border regions of the brain and form a potential connection with cell populations within the parenchyma. The fact that most adaptive immune cells of the central nervous system under homeostatic conditions are at its borders provides the brain and spinal cord an arrangement to detect and respond to perturbations while at the same time maintaining neuronal function via immunological protection (Rustenhoven and Kipnis, 2019) (Figure 1a).

Figure 1. Interplay between innate and adaptive immunity in the homoeostatic and AD brain.

(a) Under homeostatic conditions, the BBB, CSF/ISF fluid flow, and the lymphatic system work together to maintain a relatively immune privileged brain with surveilling microglia in the parenchyma with intact adaptive and innate immune system components at the border zone, which is believed to have an essential role to safeguard the brain including normal synapse and circuit functions. (b) In AD, sequential Aβ deposition and hyperphosphorylated tau accumulation induce parenchymal innate immune activation. These pathological changes affect the integrity of BBB, CSF/ISF flow and lymphatic drainage that subsequently lead to an evolving detrimental immune niche both in the brain parenchyma and border zone, including expansion of IFN responsive and antigen presenting microglia, increased inflammatory cytokine and antigen accumulation and T cell parenchymal infiltration, activation and TCR clonal expansion. These changes in the innate and adaptive immune system and their responses would therefore serve as a foundation for therapeutic development for AD.

Under disease conditions that occur in disease states such as neurodegenerative diseases, a series of complex changes occur in both innate and adaptive immunity (Dulken et al., 2019). Alzheimer’s disease (AD) is the most common neurodegenerative disease. It has been over a century, since, in 1907, Alois Alzheimer published a case report of a 51-year-old woman who exhibited symptoms of memory loss, confusion and disorientation (Alzheimer, 1907). Pathologically, AD is characterized by regional brain atrophy with corresponding neuronal and synaptic loss which together likely account for the different cognitive changes (Holtzman et al., 2011). It has also been revealed that AD is a disorder of protein aggregation in which two main proteins are involved (Duyckaerts et al., 2009). Amyloid-β (Aβ) accumulates in extracellular parenchymal amyloid plaques as well as in the walls of the cerebrovasculature to form cerebral amyloid angiopathy (CAA) (Hardy and Selkoe, 2002; Smith and Greenberg, 2009). The microtubule associated protein tau aggregates predominantly in the cytoplasm of neurons and their processes in a post-translationally modified form that includes hyperphosphorylation and other post-translational modifications in neurofibrillary tangles, dystrophic neurites, and neuropil threads (Serrano-Pozo et al., 2011; Wesseling et al., 2020). AD is more than simply a proteopathy, as Aβ and tau aggregation leads subsequently to the damage of synapses, neuronal processes, and blood-brain barrier (BBB), which in turn lead to reactive microglia and astrocytes as well as infiltration of some peripheral immune cells into the brain. Thus, even in what are not thought of as autoimmune diseases, the “immune-privileged” milieu is not immutable and both local environmental changes and immune cells infiltrating from blood and from the border regions can result in stark changes in the infrastructure of both the innate and adaptive immune system within the brain parenchyma in AD (Figure 1b).

In this review, we will focus on recent progress in the time course of the pathological characteristics of AD, advances in research regarding local innate and adaptive immunity in AD, the likely role of the immune responses in disease progression, and challenges and opportunities to develop therapeutic strategies.

AD pathology development and progression

There is a long pre-clinical phase of AD that is followed by a clinically symptomatic phase. Aβ deposition begins in parts of the neocortex and slowly and steadily accumulates in the pre-clinical (asymptomatic) phase of the disease. This process begins ~20 years prior to the onset of clinically detectable cognitive impairments (Long and Holtzman, 2019). By the time of symptom onset which usually is characterized by episodic memory impairment and executive dysfunction, Aβ deposition is already very substantial and close to its maximal extent. Tau pathology does not start to accumulate in specific regions of the neocortex until just a few years prior to symptom onset and the location of its presence is highly correlated with the onset and progression of clinically detectable cognitive impairments as well as brain atrophy (La Joie et al., 2020). Detection of these changes in protein aggregation can now be detected in the living human brain through the use of plasma, CSF, and neuroimaging biomarkers which can be of value in staging the disease, predicting outcome, and in assessing for modulation by different treatments (Barthelemy et al., 2020; Fagan et al., 2006; Fagan et al., 2007; Franzmeier et al., 2020; Nakamura et al., 2018; Ovod et al., 2017; Therriault et al., 2022). For example, as amyloid deposition begins, CSF and plasma levels of soluble, monomeric Aβ42 and the Aβ42/40 ratio decrease, which is followed by increases of different species of p-tau and total tau which together mark increasing amyloid deposition. As buildup of amyloid becomes extensive, tau pathology begins to move into the neocortex and spread and these changes can be detected with amyloid and tau imaging by PET scan (Therriault et al., 2022). As tau pathology spreads, microglial activation can be detected in vivo with PET scan using compounds such as PBR28 (Dani et al., 2018; Pascoal et al., 2021).

As the main component of amyloid plaques, Aβ is generated by sequential proteolytic cleavage of amyloid precursor protein (APP) by α- or β- secretase (BACE1) and γ-secretase in a non-amyloidogenic or amyloidogenic pattern, respectively (Goedert and Spillantini, 2006). After processing in endosomes, Aβ is then secreted into the interstitial fluid (ISF) in a process regulated by synaptic activity (Cirrito et al., 2005). Under non-pathological conditions, the predominant form of Aβ is 40 amino acids in length (Aβ40) peptide with smaller amounts of other forms of Aβ peptides including longer more amyloidogenic forms including Aβ42. In autosomal dominant AD, the early onset of Aβ aggregation appears to be driven by one of several factors depending on the gene mutation including 1) a relative overproduction of Aβ42 and longer Aβ isoforms vs. shorter forms; 2) overproduction of all forms of Aβ, or 3) a change in Aβ sequence leading to a form more prone to aggregate (Musiek and Holtzman, 2015). In late-onset AD, most evidence points to a problem with Aβ clearance (Mawuenyega et al., 2010). When Aβ does aggregate, longer forms such as Aβ42 are prone to form higher-order oligomers, protofibrils or fibrils, which are insoluble and begin to form in the brain parenchyma. Once amyloid plaques form, reactive microglia and astrocytes surround the fibrillar plaques (Condello et al., 2015). This process is linked to the formation of swollen, distorted axons and dendrites (dystrophic neurites) (Tsai et al., 2004). Notably, Aβ peptide not only accumulates in parenchymal extracellular plaques but also in the walls of the leptomeningeal and penetrating cerebrovasculature in the form of CAA. CAA often leads to blood-brain barrier (BBB) breakdown, vasculature dysregulation, ischemia, and sometimes hemorrhages (Greenberg et al., 2020).

As aforementioned, tau pathology, but not Aβ accumulation, strongly correlates with cognitive impairment and regional brain atrophy (Giannakopoulos et al., 2003). Tau is a microtubule binding protein that is predominately localized to axons of neurons under normal conditions. In AD, tau becomes hyperphosphorylated and post-translationally modified in other ways (e.g., acetylated, ubiquitinated) and aggregates into oligomers and paired helical filaments that make up structures seen in dystrophic neurites surrounding Aβ plaques (NP tau), neuropil threads (NTs), and neurofibrillary tangles (NFTs) in neuronal cell bodies (Cohen et al., 2011; Wesseling et al., 2020). Post-mortem studies and, more recently, molecular live imaging of tau with PET scans, reveal that NFTs occur hierarchically. Clinico-neuropathologic correlation analyses revealed that tau pathology propagates throughout the AD brain over several years and is distributed in a pattern across neuroanatomically connected networks, forming the basis of Braak staging (Braak and Braak, 1991). While several patterns can be seen, the most common pattern is that tau pathology is first seen in the transentorhinal cortex (Braak stages I-II) which is seen in most people with or without AD after the age of 60. Tau pathology subsequently expands to the limbic structures of the subiculum and inferior temporal neocortex (Braak III) just a few years prior to the onset of the cognitive decline. These regions are involved with functions such as memory function formation and visual object recognition This is then followed by tau pathology appearing in the amygdala and thalamus-regions involved with emotional processing and memory (Braak IV), and finally to other regions of the neocortex (Braak V-VI) as cognitive decline progresses (Braak et al., 2006).

The prion-like and spreading features of tau aggregates are almost never seen unless there is a large amount of concomitant Aβ pathology, indicating functional interplay between Aβ and tau pathology in disease progression. The functional interplay between Aβ and tau includes, 1) how does Aβ pathology affect tau pathology and tau-associated neurodegeneration? 2) will tau pathology in turn have consequences on Aβ pathology? By injecting human AD-tau into the brains of amyloid plaque-bearing transgenic mice, which harbor Aβ plaques, all three types of tau pathology (NP tau, NTs and NFTs) were successfully generated in a time course suggesting that Aβ plaques have created a unique environment that facilitates the rapid amplification of proteopathic AD-tau seeds. Importantly, in this model system, microglia play a key role in this process. Aβ plaques are necessary but not sufficient to initiate the cascade of pathological tau transmission, as misfolded tau seeds with the correct conformation are also required (He et al., 2018). In mice expressing human Aβ and tau, tau-dependent neuronal activity suppression dominates over Aβ-dependent neuronal hyperactivity, which again indicates tau, rather than Aβ determines functional status (Busche et al., 2019; Roberson et al., 2007). A prevailing hypothesis is that there is an Aβ-dependent disease period which predominantly occurs during the pre-symptomatic phase of the disease which is necessary, but not sufficient, for the full-blown disease (Musiek and Holtzman, 2015). This period is important in determining the age of symptom onset. There is then a later more Aβ-independent period which drives disease progression (Busche and Hyman, 2020). If this hypothesis is correct, anti-Aβ therapies are likely to be of much less benefit if the intervention is pursued in the Aβ-independent disease period. Considering the timing of these events and the functional interplay between Aβ and tau, clinical trials targeting Aβ before or during an Aβ-dependent disease period and targeting tau or combination therapies in the Aβ-independent disease period would be expected to result in more promising outcomes. Meanwhile, delineating the underlying driving force and functional interplay between Aβ and Tau, including changes in the innate and adaptive immune response, lipid metabolism, autophagy, proteasome function, vesicle transport and mitochondrial function, might pave a way to modulate the period of disease evolution and ultimately disease progression.

AD genetic risk factors and their implications on innate immunity and AD progression

Rare forms of autosomal dominant AD are caused by mutation in one of 3 genes, β-amyloid precursor protein (APP), presenilin 1 (PSEN1), or presenilin 2 (PSEN2) (Goate et al., 1991; Tanzi, 2012). Clinical onset of disease in these cases is usually between the ages of 30–60. Late-onset AD (LOAD), age of onset after age 65, is also highly influenced by genetics and genetic studies have been of great importance for understanding the underlying etiology of LOAD. Genome-wide associated studies (GWAS) as well as other methods have identified a plethora of AD risk factors, and intriguingly more than half of the risk loci that clearly involve a specific gene are significantly enriched or uniquely expressed in immune cells, especially microglia and macrophages, including APOE, triggering receptor expressed on myeloid cells 2 (TREM2), ATP-binding cassette transporter (ABCA) family, Complement, CD33, HLA-family, MEF2C, and MS4A family (Griciuc and Tanzi, 2021; Karch and Goate, 2015; Pimenova et al., 2018). Modeling these genetic variations in animals under basal conditions or in disease relevant contexts has revealed that immune responses are not just a secondary response to AD pathology but are also a key driving factor for disease development and progression. How AD genetic risk factors link to innate immunity and their function in the pathogenesis of AD is discussed in the following sections.

ApoE is an apolipoprotein that is present in lipoprotein particles. Depending on the type of lipoprotein it is present in (e.g. VLDL, LDL, HDL), it can transport lipids and cholesterol to target cells via receptor-mediated endocytosis and when in HDL also participate in reverse transport for lipids and cholesterol efflux out of the cells (Mahley, 1988). ApoE is expressed in many organs with very high protein levels in liver and brain. In the brain, ApoE is primarily produced by astrocytes and oligodendrocytes under homeostatic conditions, while its expression is greatly increased in reactive microglia under disease conditions. Low-density lipoprotein receptor (Ldlr) and lipoprotein receptor-related protein 1 (Lrp1) are major metabolic receptors in mediating the levels of and other effects of ApoE in the brain (Holtzman et al., 2012). In terms of its function in mediating cellular cholesterol and lipid efflux in the CNS, ApoE-containing HDL participates in this process through increasing its lipidation state via Abca1 and Abcg1 (Hirsch-Reinshagen et al., 2004; Wahrle et al., 2004). Human APOE is a 299 amino acid protein, and has three variants, E2 (cys112/cys158), E3 (cys112/arg158) and E4 (arg112/arg158) (Martens et al., 2022). Numerous studies have revealed that APOE is the strongest genetic risk factor for LOAD (Kim et al., 2009). A recent case report showed that an individual who had two copies of the APOE3 Christchurch (R136S) mutation was relatively resistant to cognitive decline due to autosomal dominant AD (Arboleda-Velasquez et al., 2019). The mechanism underlying this effect is not clear. However, the individual had large amounts of amyloid pathology and less tau pathology with a different spatial localization than other cases suggesting that this APOE variant is acting at the interface between Aβ and tau (Sepulveda-Falla et al., 2022), possibly via influencing the innate immune response. In another case, the APOE3 variant (V236E) appears to result in decreased Aβ pathology by reducing APOE4 aggregation and increasing cholesterol efflux (Liu et al., 2021). The ε4 allele (encoding APOE4) increases and the ε2 allele decreases disease risk in a dose-dependent fashion relative to the ε3 allele (Corder et al., 1993). APOE4 carriers have more amyloid deposition, earlier disease onset, and progress more rapidly once the symptomatic phase initiates (Huynh et al., 2017a). Transgenic animal studies have recapitulated clinical observations and have confirmed that ApoE strongly affects Aβ deposition, tau-mediated neurodegeneration, and other phenotypes such as BBB dysfunction, antigen presentation, and T cell activation (Bonacina et al., 2018; Montagne et al., 2020) (see Table 1. AD mouse models, their pathogenesis and potential immune interactions). In terms of its role in brain lipid metabolism, ApoE has been shown to influence both astrocyte and microglial lipid metabolism, which may be relevant to some of its effects in AD pathology (Nugent et al., 2020; Tcw et al., 2022).

Table 1.

AD mouse models, their pathogenesis and potential immune interactions.

Mouse models	Genetic mutations	Pathogenesis	Potential immune interactions
5xFAD	K670N/M671L (Swedish), I716V (Florida), V717I (London) in APP, and the M146L and L286V in PSEN1	An Aß mouse model with amyloid plaques, accompanied by microgliosis and astrogliosis at around 2 months of age. Pathology is more severe in females than that in males before plateau at around 10 months.	1. Trem2 KO or R47H mutation decreases plaque associated microglia and increases surrounding dystrophic neurites (referring to Trem2, Song et al., 2018; Wang et al., 2016) 2. Blockage of IFNAR diminishes microglial activation and engulfment (referring to Interferon signaling, Roy et al., 2020; Roy et al., 2022). 3. Enhancement of lymphatic function promotes Aß clearance (referring to Meninges and dural sinus, Da Mesquita et al., 2018; Da Mesquita et al., 2021). 4. Genetic ablation of peripheral immune cell populations through Rag KO accelerates Aβ pathology, enhances neuroinflammation, increases microglial activation, and decreases microglial phagocytosis (referring to Adaptive immunity and AD, Marsh et al., 2016). 5. Depletion of Treg mitigates Aß pathology (referring to Adaptive immunity and AD, Baruch et al., 2015). 6. Depletion of activated B cells by anti-CD20/B220 antibody ameliorates Aβ pathology (referring to Aaptive immunity and AD, Kim et al., 2021).
APP/PS1–21	K670N/M671L (Swedish) in APP and L166P in PSEN1	An Aß mouse model with amyloid plaques, accompanied by microgliosis and astrogliosis at around 6 weeks of age in the cortex, followed by deposits in hippocampus at around 3–4 months.	1. ApoE KO or selective removal from astrocytes decreases Aß pathology (referring to ApoE, Ulrich et al., 2018; Mahan et al., 2022). 2. Blockage of IFN-γ reduces microglia activation and plaque burden (referring to Iterferon signaling, Browne et al., 2013).
APP/PS1∆E9	K670N/M671L (Swedish) in APP, exon 9 deletion in PSEN1	An Aß mouse model with amyloid plaques, accompanied by microgliosis and astrogliosis at around 6 months of age.	1. Inhibition of complement pathway reduces phagocytic microglia and synaptic loss (referring to Complement, Lian et al., 2016). 2. KO of NLRP3 or caspase-1 diminishes Aß plaque formation and increases microglial phagocytosis (referring to NLRP3 inflammasome pathway, Halle et al., 2008; Heneka et al., 2013; Venegas et al., 2017). 3. TAM loss-of-function leads to deficient Aß phagocytosis in microglia (referring to TAM signaling, Huang et al., 2021). 4. Depletion of Ly6C-low monocytes induces Aß deposition in brain parenchyma (referring to other innate immune cells in AD, Michaud et al., 2013). 5. Genetic ablation of peripheral immune cell populations through Rag2 KO reduces fibrillar Aβ deposits and total Aβ, while Rag2 KO bone barrow reconstitution enhances microgliosis and Aβ phagocytosis (referring to Adaptive immunity in AD, Spani et al., 2015).
P301S	P301S in MAPT	A Tauopathy mouse model with neurofibrillary Tau tangle pathology, microgliosis, astrogliosis at around 5–6 months of age, and synaptic and neuronal loss at around 9 months of age.	1. ApoE KO or selective removal from astrocytes decreases Tau pathology (referring to ApoE, Wang et al., 2021). 2 Inhibition of complement pathway attenuates microglia activation and Tauopathy (referring to Complement, Litvinchuk et al., 2018). 3. Inactivation of NF-κB partially restores microglial homeostasis and Tau-mediated cognitive decline (referring to NF-κB signaling, Wang et al., 2022). 4. Microglia depletion attenuates Tauopathy (referring to Innate immunity and microglia, Mancuso et al., 2019; Shi et al., 2019).
3xTg-AD	K670N/M671L (Swedish) in APP, M146V in PSEN1 and P301L in MAPT	An Aß and Tauopathy mouse model with amyloid plaques at around 6 months of age and Tauopathy later.	1. Neutrophil depletion or blockage of LFA-1 reduces plaque pathology and memory decline (referring to Other innate immune cells in AD, Zenaro et al., 2015). 2. Deficiency in B cells reduces Aß burden and microglial activation (referring to Adpative immunity and AD, Kim et al., 2021).
Tau22	G272V and P301S in MAPT	A Tauopathy mouse model with neurofibrillary Tau tangle pathology, microgliosis and astrogliosis at around 5–6 months of age.	KO of NLRP3 or ASC reduces Tau hyperphosphorylation and aggregation, and microglial inflammation (referring to NLRP3 inflammasome pathway, Ising et al., 2019).

Abundant data on innate immune responses, especially the role of microglia in AD pathogenesis have been studied. Whether and how adaptive immune responses interact with innate immune components in AD pathogenesis, and their functional consequences on disease progression require further studies.

ApoE is essential for Aβ seeding and plaque formation. When Aβ depositing mice were crossed to ApoE^−/− mice, little to no fibrillar Aβ or CAA ever formed (Bales et al., 1997; Holtzman et al., 2000). Reduction of ApoE levels with antisense oligonucleotides (ASOs) from just after birth led to a significant decrease in Aβ seeding and deposition and decreased Aβ pathology at 4 months of age in APP/PS1-21 mice expressing human APOE3 or APOE4 (Huynh et al., 2017b). In contrast, overexpression of APOE4 from birth in APP/PS1 transgenic mice increased Aβ seeding and pathology (Liu et al., 2017a). Most of the effects of APOE on Aβ seeding and plaque formation appears to be derived from astrocytes (Mahan et al., 2022). Selective knockout of murine ApoE from microglia by Csf1r-Cre/ApoE ^f/f had no major effect on Aβ deposition in the 5xFAD mouse model (Henningfield et al., 2022); however, selective removal of human APOE isoforms from microglia by expressing Cre using a microglial-specific promoter in APOE^f/f mice and the effect of that manipulation on Aβ-related pathologies remains to be assessed. Interestingly, endogenous ApoE^−/− in aggressive Aβ-depositing mice such as APP/PS1-21 and APP/PS1ΔE9 mice led to strongly reduced fibrillar Aβ and increased non-fibrillar Aβ deposits. Around the few fibrillar plaques that did form in the brain of ApoE^−/− mice, reduced plaque-associated microgliosis and more severe neuritic dystrophy were observed, pointing to an important role of ApoE in mediating microglia activation and plaque compaction, and shaping local Aβ induced neural process injury (Ulrich et al., 2018). On the other hand, ApoE competitively bound to its receptors that are also receptors for Aβ, such as Ldlr and Lrp1 on astrocytes, microglia and endothelial cells, and therefore impaired Aβ clearance (Verghese et al., 2013). Whether Aβ clearance by glymphatic and meningeal lymphatics is differentially regulated by APOE isoforms remains unknown and needs to be further studied.

In tau-mediated neurodegeneration, ApoE also has important roles and shows a differential effect of isoforms. In sporadic tauopathies, it has been reported that individuals bearing an APOE4 allele had more neurodegeneration in the presence similar amounts of tau pathology (Shi et al., 2017). In P301S tau transgenic mice crossed to human APOE knockin mice, higher tau pathology and markedly more brain atrophy as well as reactive microglia and astrocytes were observed in the presence of APOE4 compared to APOE2 or APOE3 (Shi et al., 2017; Yoshiyama et al., 2007). In addition, endogenous ApoE^−/− or selectively removal of astrocytic APOE4 in P301S mice resulted in diminished microglial- and astrocyte-mediated reactivity, disease-associated gene expression changes, and regional brain atrophy (Wang et al., 2021). Whether selective removal of APOE4 from microglia will influence tau pathology or tau-mediated neurodegeneration remains to be answered.

Trem2 is a member of the immunoglobulin superfamily of receptors that is expressed in osteoclasts, macrophages and microglia, emphasizing its potential role in phagocytosis, metabolic regulation and immune modulation (Nugent et al., 2020). Phospholipids, lipoproteins including ApoE, Aβ oligomers, and apoptotic neurons have all been reported as potential Trem2 ligands (Atagi et al., 2015; Krasemann et al., 2017; Ulrich et al., 2017; Wang et al., 2015; Zhao et al., 2018). Single amino acid differences at specific sites such as TREM2 R47H in humans are associated with 2–4 fold increased risk for AD and these mutations are linked with decreased TREM2 function (Guerreiro et al., 2013; Jonsson et al., 2013). After ligand binding to the receptor extracellular domain, the TREM2 transmembrane domain associates with tyrosine kinase-binding protein (DAP12) and hematopoietic cell signal transducer (DAP10), which in turn recruit tyrosine-protein kinase (SYK) and phosphatidylinositol 3-kinase (PI3K) for downstream signaling, respectively. Downstream pathways of TREM2-DAP12/DAP10 signaling include intracellular Ca²⁺ mobilization, activation of mitogen-activated protein kinase (MAPK) and mTOR signaling (Bouchon et al., 2001; Ulland and Colonna, 2018). Soluble TREM2 (sTREM2), the ectodomain of TREM2, is released upon shedding by ADAM10/17 (a disintegrin and metalloproteinase domain-containing protein) cleavage (Kleinberger et al., 2014; Wunderlich et al., 2013). The translation of preclinical findings on TREM2 has been facilitated by development of sTREM2 as a biomarker in CSF for microglial mediated inflammation and neuronal injury in AD patients (Suarez-Calvet et al., 2016). Increased sTREM2 was associated with memory decline and hippocampal shrinkage in patients with AD (Ewers et al., 2019). sTREM2 was positively correlated with CSF tau and p-tau and associated with genetic variant carriers, such as R47H (Piccio et al., 2016). sTREM2 promoted microglia survival and triggered microglia mediated inflammatory cytokine response in mice (Zhong et al., 2017).

Similar to ApoE, Trem2 shows potential dual roles for disease-associated microglia and disease progression in the context of amyloid plaque formation and tau pathology-mediated brain atrophy. In 5xFAD amyloid depositing mice, Trem2^−/− or the TREM2 R47H mutation linked with increased AD risk resulted in a decrease in plaque associated microglia and an increase in surrounding dystrophic neurites (Song et al., 2018; Wang et al., 2016). Loss of plaque-associated microglia in the presence of reduced TREM2 function or Trem2^−/− has also been shown to significantly exacerbate Aβ-associated tau seeding and spreading in amyloid models (Gratuze et al., 2021; Leyns et al., 2019; Ulrich et al., 2014). As decreased Trem2 function has been linked to increased Aβ associated local toxicity and Aβ-linked tau seeding and spreading in the brain, several companies have developed treatments including antibodies that activate TREM2 which have entered clinical trials in AD. In amyloid depositing mice, fibrillar Aβ deposits were significantly elevated in 3 month old Trem2^−/− mice with a subsequent lower accumulation rate of fibrillar Aβ during ageing, suggesting a role of Trem2 in restricting amyloid deposition at an early stage (Parhizkar et al., 2019) and that an age-dependent local microenvironment might convert Trem2 from beneficial to having more detrimental effects during disease progression. In contrast to the effect of Trem2 in Aβ models, in a mouse model of tauopathy (P301S mice), Trem2^−/− or the TREM2 R47H mutation attenuated microglia reactivity and protected against brain atrophy (Gratuze et al., 2020; Leyns et al., 2017; Sayed et al., 2018). This suggests a beneficial role of Trem2 activation in the setting of Aβ pathology but its activation may be otherwise detrimental when more severe neurodegeneration proceeds. Further studies in this area are clearly needed to better understand the effects of Trem2 activation in modulating the function of microglia either acutely or chronically in AD as well as the translational implications of these findings.

Microglia and astrocytes are the cells in the CNS that synthesize and secrete most complement proteins and express their receptors. In the context of aging and AD, complement pathways and their clearance functions are elevated. C1q, the initiating protein of the classical complement cascade, was increased and associated with synapses before overt plaque deposition in the Aβ depositing mouse model J20, which appeared to contribute to synaptic loss (Hong et al., 2016). Inhibition of C1q, C3, or the microglial complement receptor Cr3 in APP/PS1ΔE9 mice, reduced the number of phagocytic microglia as well as the extent of early synaptic loss (Lian et al., 2016). In a model of tauopathy, C3ar1^−/− resulted in attenuation of neuroinflammation, synaptic deficits, tau pathology as well as neurodegeneration in P301S tau mice (Litvinchuk et al., 2018). Treatment with C5 siRNA attenuated Aβ-associated microglial accumulation in APP/PS1-21 mice. In addition to binding to synapses to refine synapse and circuit wiring, C1q also binds to Aβ and ApoE directly. C1q-ApoE complexes were observed in choroid plexus and Aβ plaques in AD patients and also in APP/PS1-21 mice (Yin et al., 2019). These results suggest that the aberrantly activated complement pathways are also a hallmark and driving force of both Aβ pathology and of tau-mediated neurodegeneration and attenuation of different aspects of the overactivated complement pathways is a potential therapeutic target for limiting neurodegeneration-associated pathology in AD.

Key immune-related signaling pathways in innate immunity in AD

Extensive studies have highlighted the importance of ApoE, Trem2, and the complement pathway in aspects of AD pathological development and progression. There is increasing evidence that immune signaling pathways are also prominent in AD-associated immune cellular changes and play potentially important roles in AD pathology.

Nuclear factor kappa-light-chain-enhancer of activated B cells (NF-κB) is a transcription factor known to modulate many target genes that are associated with neuroinflammation, glial activation, oxidative stress, cell proliferation, and apoptosis in the brain (Liu et al., 2017b). NF-κB was activated in primary neurons after Aβ peptide treatment and in neurons surrounding early plaques in the brain of AD patients (Kaltschmidt et al., 1997). Tau pathology also activates NF-κB signaling in microglia. Inactivation of NF-κB partially restored microglial homeostasis, reversed tau-mediated transcriptional changes and rescued spatial learning and memory deficits in P301S mice tau transgenic mice, suggesting a detrimental role of overactive NF-κB signaling in tau-mediated brain injury (Wang et al., 2022).

NOD-, LRR- and pyrin domain-containing protein (NLRP3) is predominantly expressed in macrophages and activated microglia (Heneka et al., 2018). NLRP3, apoptosis-associated speck-like protein (ASC) and pro-Caspase-1 forms the NLRP3 inflammasome that initiates the production of the proinflammatory cytokine, interleukin-1β (IL-1β) and IL-18. NLRP3^−/− or Caspase-1^−/− strongly diminished Aβ plaque formation, increased microglial-mediated Aβ phagocytosis and improved spatial memory in 16-month-old APP/PS1ΔE9 Aβ depositing mice (Halle et al., 2008; Heneka et al., 2013; Venegas et al., 2017). The NLRP3 inflammasome was also activated in patients with primary tauopathies and mice with tau pathology. Nlrp3^−/− or Asc^−/− reduced tau hyperphosphorylation and aggregation in 8 month old Tau22 mice, as well as Aβ-induced tau pathology, suggesting that microglial inflammation may be one of the factors that triggers tau pathology downstream of Aβ pathology (Ising et al., 2019).

mTOR is a serine/threonine protein kinase, which is involved in regulation of protein synthesis, cellular energetics, and biosynthetic metabolism, degradation, and autophagy (Saxton and Sabatini, 2017). mTOR is widely expressed in neurons, astrocytes and microglia, and interacts with several other proteins to form two distinct complexes, mTORC1 and mTORC2. mTOR activity is tightly regulated and basal autophagy levels are maintained in homeostatic brain. Impaired mTOR activation and microglial apoptosis, associated with low energy states and enhanced compensatory autophagy in response to metabolic stress, were observed in microglia in 5xFAD amyloid depositing mice with Trem2^−/−, indicating Trem2 as an innate immune receptor that impacts microglial metabolism in the presence of AD pathology through the mTOR pathway. Addition of cyclocreatine increased microglial mTOR signaling and cell viability, rescued energetic microglial metabolism, and provided protective responses against Aβ associated neurite dystrophy in 5xFAD mice with Trem2^−/−, suggesting mTOR-mediated microglial metabolism may be a fundamentally distinct therapeutic approach (Ulland et al., 2017).

TAM signaling is driven by activation of the Tyro3, Axl, and Mer tyrosine kinase receptors by their ligands Gas6 and Protein S (Rothlin et al., 2015). Axl and Mer are expressed in microglia and macrophages and are required for their phagocytosis function. Axl and Mer are even more prominently upregulated in amyloid plaque-associated microglia in 7–15 month old APP/PS1ΔE9 mice (Huang et al., 2021). TAM^−/− mice developed less dense-core plaques and increased CAA, indicating TAM-deficient microglia have deficient Aβ phagocytosis and altered parenchymal plaque and CAA formation.

Interferons (IFNs) are widely expressed cytokines that have classical antiviral and immunosurveillance effects. The IFN family includes two main classes, IFN-I and IFN-II (IFN-γ) (Platanias, 2005). IFN-I binding to IFNAR1 and IFNAR2 activates the JAK-STAT signaling pathway and initiates the transcription of a large panel of IFN-stimulated genes (ISGs), including Isg15, Ifitm1/2/3, Irf1/7 and Mx1 (McNab et al., 2015). In amyloid depositing AD models, activated ISG-expressing microglia exclusively surrounded Aβ plaques containing nucleic acids in an age-dependent manner. Blockade of IFNARs effectively diminished Stat1⁺, CD68⁺, Clelc7a⁺ microglial numbers and synapse loss in 3 month old 5xFAD mice (Roy et al., 2020). Microglia-specific IFNAR1 deletion attenuated cognitive defects and engulfment of post-synaptic terminals in 5-month-old 5xFAD mice (Roy et al., 2022). Recent studies also reveal that CNS-infiltrating T cells, especially T helper (Th1) cells expressed IFN-γ, and blockade of IFN-γ signaling via neutralizing antibodies reduced CD11b expression and plaque burden in 6–7 month old APP/PS1 mice (Browne et al., 2013). Of note, the innate immune receptors and AD-associated triggers such as amyloid as well as how IFN production in AD is coordinated still remains to be determined. As tau accumulation and tau-mediated neurodegeneration are strongly linked to cognitive and functional decline in AD and primary tauopathies, future studies are needed to address the core question of whether IFN signaling drives neuroinflammation and neurodegeneration in the setting of tau pathology.

Immune system and immune responses in the brain

The immunological defenses in vertebrates consist of two subsystems: the innate and adaptive immune responses (Vivier and Malissen, 2005). The innate immune system responds to pathogens by germline-encoded receptors, which are expressed particularly in cell types not clonally distributed. During the innate immune response, pattern-recognition receptors, like Toll like receptors (TLRs), recognize microbial pathogens and damage associated molecular patterns (misfolded/aggregated proteins, denatured DNA and LPS) and are essential for triggering activation of NF-κB signaling and inflammation (Medzhitov, 2001). The cellular components of the innate immune system include dendritic cells (DCs), monocytes, macrophages (and also microglia in brain) and natural killer T cells. Innate immunity responds rapidly to triggers but its responses are relatively nonspecific. In contrast, the adaptive immune responses can be very specific. The specificity depends on the large repertories of antigen receptors on T and B cells (TCR and BCR), the cellular elements of the adaptive immune system (Clark and Kupper, 2005). TCR and BCR diversity is generated somatically through site-specific DNA recombination and each receptor of a particular specificity is expressed in a clonal way. Together these receptors make up a recognition repertoire that increases the possibility that the adaptive immune response can detect many possible antigens encountered throughout life. Therefore, the nature of innate and adaptive repertoire constitutes a key difference between these two types of immunity.

Consistent with the two cellular composition arms of the immune systems, the immune responses are divided into afferent and efferent arms. The afferent arm of the adaptive immune response involves antigen presentation to naïve T cells, resulting in their priming and activation. Soluble antigens and immune cells carrying antigens from the brain are transported to the deep cervical lymph nodes (dCLN), where those antigens are presented to naïve T cells and B cells by professional antigen presenting cells (APCs), such as mature dendritic cells (Papadopoulos et al., 2020; Santambrogio et al., 1999). Antigen-specific T cells and B cells become activated and undergo clonal expansion after trafficking back into the brain. Under disease conditions, APCs accumulate in the inflamed brain (Garber et al., 2019). It is still not clear whether endogenous or infiltrated APCs in the brain parenchyma function as antigen presenting in situ and lead to an adaptive immune response. The entry of new APCs, T cells, and B cells into the brain as well as the local response of resident microglia to local disease pathology result in an altered immune state in the brain parenchyma differing from that in the homeostatic brain. The efferent arm of the adaptive immune response starts when antigen-specific T cells reach the brain parenchyma and exert their effector function. However, before that, these cells need to pass either across the BBB or from brain borders to the parenchyma where they newly encounter a unique brain niche, including neurons, microglia, astrocytes, oligodendrocytes, endothelial cells, extracellular matrix, and brain fluids. Both the immune system and the central nervous system continuously surveil the environment and make on-demand adjustments to maintain homeostasis.

Immune niche safeguards in the brain and their dysregulations in AD

Both the innate and adaptive immune responses are present in the CNS, although the immune niche is safeguarded as “immune privileged” in the homeostatic brain parenchyma. In neurodegenerative diseases including AD, abundant data have emerged regarding the involvement of both the innate and adaptive immune responses in the initiation or progression of the disease, indicative of a remodeling of the immune state and the two arms of immunity within the brain parenchyma. Here, we review recent progresses on boarder structures safeguarding the immune niche and their changes, which might transform the innate and adaptive immune system during the course of AD development and progression in humans and model systems (Figure 1a and b) and provide our perspectives for future studies.

The Blood-brain barrier (BBB) is made up of a continuous endothelial cell barrier within brain microvessels that has sealed cell-to-cell contacts and is sheathed by pericytes and astrocyte end-feet (Daneman and Prat, 2015). The unique structures of brain capillaries result in high transendothelial electrical resistance and low paracellular and transcellular permeability to restrict the entry of blood carried molecules into the brain, unless they have specialized receptors and transporters in the endothelial cells. Neuroimaging and pathological studies have demonstrated BBB dysfunction in AD and other neurodegenerative diseases as well as in models with AD pathology (Montagne et al., 2020). These changes include increased BBB permeability, microhemorrhages, reduced glucose transport, impaired P-glycoprotein-1 efflux function, reduction of Aβ clearance across the BBB, and increased leukocyte infiltration (Sweeney et al., 2018). Histological studies have also revealed profound pathological changes suggesting BBB degeneration in post-mortem brain tissue of AD patients, in vitro human BBB-like structures derived from induced pluripotent stem cells from AD patients, and transgenic mouse models with AD pathologies (Blanchard et al., 2020). These pathological changes include blood-derived protein leakage, pericyte and endothelial cell degeneration, and peripheral cell infiltration. While studies regarding the dynamic cellular and molecular changes of the BBB in pre-clinical and the clinically symptomatic phase of AD have been done, more studies of the regional specificity of changes to the BBB and its implications in remodeling regional immunity and therefore magnifying AD progression are required. This will allow for assessments of potential therapeutics to modulate the BBB both to restore its function during neurodegeneration and to potentially develop methods to bypass the BBB for drug delivery such as has been recently been accomplished with the use of transferrin receptor binding biologics (Logan et al., 2021).

The meninges are a triple-layer structure enveloping the brain, containing two inner thin layers, pia and arachnoid matter (together, leptomeninges) and the outer dural layer (Da Mesquita et al., 2018a). CSF is produced by specialized epithelial cells of the choroid plexus and flows in the sub-arachnoid space within the leptomeninges. Data suggest that CSF can flow around perivascular spaces into the interstitial fluid (ISF) of the brain and this flow can be modulated by aquaporin-4 (AQP4) water channels present in the astrocyte end-feed, the so called glymphatic system. AQP4 polarized distribution is observed in the perivascular end feet of astrocytes, while defective expression and mis-localization lead to the reduction of Aβ clearance which is relevant to AD (Iliff et al., 2012). It will be interesting to determine in future studies the role of perivascular macrophages which surround CNS penetrating vessels in CSF/ISF flow. This flow is likely important in exchange of extracellular molecules in the parenchyma, such as cytokines, soluble Aβ, and antigens within the ISF which can then re-enter CSF before exiting the CNS via the dural lymphatics.

Single cell transcriptomics studies have successfully delineated the immune populations within the dura and dural sinus, which includes macrophages, DCs, innate lymphoid cells (ILCs), mast cells, neutrophils, B cells, and T cells (Rustenhoven et al., 2021). T cells are more likely blood derived, while B cells, monocytes and neutrophils appear to derive from adjacent bone marrow from skull to brain channels (Cugurra et al., 2021). CNS-derived antigens in the CSF accumulate around the dural sinuses, are captured by local APCs, and can be presented to patrolling T cells. This surveillance is enabled by endothelial and mural cells forming the sinus stromal niche. T cell recognition of DC- and macrophage-presented antigens at this site promotes tissue resident phenotypes and effector functions within the dural meninges.

Interestingly, functional lymphatic vessels are present within the dura and are able to carry fluid and cytokines from CSF exiting from the brain parenchyma as well as local immune cells, such as T cells, to the dCLN, allowing interactions between brain derived antigens and peripheral T cells (Ahn et al., 2019; Aspelund et al., 2015; Louveau et al., 2015b). The discovery of the dural lymphatic system has led to new insights into how the CNS, its border zones, and the periphery can communicate, especially in regard to neuroimmune interactions. This may shed important new light on neuroimmune players in neurodegenerative diseases. In addition to providing a route for molecules to reach into and drain out of the brain independent of the BBB, the meninges also provide a location where peripheral immune cells can sample the soluble antigenic repertoires immediately presented by APCs to T cells locally even before passage to the dCLNs (Rustenhoven et al., 2021).

Meningeal lymphatic dysfunction was observed in 13–14 month old aged 5xFAD Aβ depositing mice which is accompanied by a significant increase in Aβ deposition throughout the meninges. Ablation of meningeal lymphatics in 5xFAD mice exacerbated Aβ deposition in the brain and meninges, microgliosis and neurovascular dysfunction, and affected the outcome of anti-Aβ immunotherapy. Interestingly, enhancement of meningeal lymphatic function combined with Aβ antibody immunotherapy improved Aβ removal from the brain (Da Mesquita et al., 2018b; Da Mesquita et al., 2021). In the future, it will be of great interest to address the fundamental question of how meningeal lymphatic endothelial cells and other cells that make up the meningeal lymphatic vessels in the human CNS, are influenced by AD-related pathologies concomitantly, and whether targeting the brain lymphatic system is also an efficient way to influence neuroimmune interactions to influence tauopathy and tau-mediated neurodegeneration (Figure 1b).

Reconstructed innate immune niche in the brain parenchyma in AD

Loss of safeguards including vascular inflammation, dysfunctional BBB, and impaired drainage via the meningeal lymphatic system have been observed in AD. Comprehensive signaling components, including neurotransmitters, chemokines and cytokines, antigens, classes of receptor systems induced by apoptotic neurons, demyelination, and gradient signals generated by reactive microglia and astrocytes and microvascular endothelial cells in the brain parenchyma could result in leukocyte infiltration when safeguards are dysregulated. This likely results in major changes in both innate and adaptive immunity within the brain parenchyma that can affect disease progression.

Microglia are brain resident innate immune cells, which arise from yolk sac erythromyeloid precursors and migrate into the brain parenchyma during embryonic development (Madore et al., 2020). It is reported that microglia are extensively involved in both brain development and brain functional modulation before and after birth, i.e. immune surveillance, shaping neuronal connectivity and function, phagocytosis of debris and apoptotic cells, and demyelination/remyelination (Madore et al., 2020; Poliani et al., 2015; Santos and Fields, 2021; Sierra et al., 2019). Neuroinflammation is present in the brain of individuals with AD and other neurodegenerative diseases, and numerous studies have focused on the cellular and molecular changes in microglia, a key cellular component of the innate immune response in the brain during AD development and progression (Heneka et al., 2014). Genome-wide transcriptome studies in single cells and with spatial transcriptomics in both animal models and human tissues have provided a chance to re-visit the contribution of microglia to AD pathogenesis in both a spatial and temporal way beyond the effect of genetic risk factors discussed previously (Olah et al., 2020). In the setting of amyloid deposition, disease associated microglia (DAM) (also termed MGnD) and phagocytic microglia can limit amyloid associated pathology and inhibit Aβ mediated tau seeding and spreading (Huang et al., 2021; Krasemann et al., 2017; Leyns et al., 2019). Single cell studies in microglia from cortex and hippocampus of male and female App^NL-G-F mice over time revealed a transition from DAM to additional microglial populations with activated response microglia (ARMs) and interferon response microglia (IRMs). ARMs and IRMs are parts of the normal evolution of microglia in healthy aging, while Aβ boosted the redistribution of homeostatic microglia to ARMs, which enriched genes involved in major histocompatibility complex (MHC) class II presentation (H2-Ab1, H2-Aa and CD74) and known AD risk genes such as ApoE, Trem2, DAP12, Ctsb, Spp1. ApoE^−/− reduced the number of microglia displaying ARM signatures and affected the interaction of microglia with Aβ plaques, supporting the direct immunomodulatory function of ApoE (Sala Frigerio et al., 2019). While there is some overlap in microglia gene expression profiles in human AD vs. what is seen in mouse models with AD pathology with an increase in expression of the MHC-II genes CD74 and HLA-DRB1 as well as APOE, there is not complete overlap (Mathys et al., 2019). Spatial transcriptomic studies of both the App^NL-G-F mouse model and human brain sections revealed plaque-induced genes, signals encompassing complement, endosomes and lysosomes, oxidation and inflammation, and the cellular niche related to Aβ in situ (Chen et al., 2020). In the setting of tau accumulation, reactive microglia and DAM/MGnD show featured gene expression patterns strongly associated with tau pathology and CNS injury. Depletion of microglia via inhibiting of colony-stimulating factor 1 receptor (Csf1r), essential for microglial survival, at the time window when tau-mediated neurodegeneration develops, strongly reduced brain atrophy and neuronal loss (Mancuso et al., 2019). In addition to affecting tauopathy and neuronal loss, activated microglia, including DAM and IRM, can directly secrete cytokines and likely mediate synapse destruction through several mechanisms including complement pathways. Accordingly, monitoring of microglial state transitions and developing specific compounds targeting microglia-mediated neurotoxicity and inflammatory pathways might serve as a way for staging AD and developing strategies to slow or halt AD progression (Block et al., 2007) (Figure 2).

Fig. 2. Microglial states and a model for a functional shift related to AD.

Microglia are brain resident innate immune cells, which have major physiological functions in immune surveillance, phagocytosis and shaping neuronal functions under homeostatic conditions. In the setting of amyloid deposition, homeostatic microglia are transformed into disease associated microglia by downregulation of genes such as Cx3cr1, Tmem119 and P2ry12, and upregulation of genes such as Trem2, ApoE and Itgax. With amyloid deposition and accumulation, disease associated microglia mainly exert “protective” effects by limiting amyloid associated pathology. In the setting of tau accumulation, self-antigens and debris generated by demyelination or neuronal death trigger microglial phagocytosis and the disease associated microglia have a greater response to IFN and express MHC-I/MHC-II molecules. The IFN responsive and antigen presenting microglia could then potentially communicate with the adaptive immune system, and the detrimental immune microenvironment composed of both innate and adaptive immune cells could lead to a state with accelerated machinery driving neurodegeneration and brain atrophy.

Leukocyte subpopulations including monocytes, macrophages, and neutrophils have been identified in the brain of AD animal models and humans. Neutrophils are typically the first line of innate immune cells of acute inflammation, and they also contribute to chronic inflammatory and adaptive immune responses (Kolaczkowska and Kubes, 2013). Neutrophils have been observed infiltrating into brain parenchyma and migrating to amyloid plaques in 5xFAD mice by in vivo imaging with a Ly6C/G fluorescently labelled antibody (Baik et al., 2014). Soluble oligomeric Aβ triggered a LFA-1 integrin- and ICAM-1 dose-dependent infiltration of neutrophils in 5xFAD and 3xTg-AD mouse brain. Neutrophil depletion or blockade of LFA-1 integrin reduced plaque pathology and memory decline in 3xTg-AD mice (Zenaro et al., 2015). Monocytes and macrophages are bone marrow-derived myeloid cells that circulate in the blood or populate in tissues (Hettinger et al., 2013). Based on CCR2 and Ly6C expression, proinflammatory monocytes (CCR2^highLy6C^high) were recruited into the tissue upon inflammatory cues and anti-inflammatory monocytes (CCR2^lowLy6C^low) patrolled the blood vessel lumen (Ginhoux and Jung, 2014). In vivo imaging studies revealed that Ly6C^low monocytes were attracted to and crawled onto the luminal walls of Aβ positive veins. Selective depletion of Ly6C^low monocytes in APP/PS1 mice induced Aβ deposition (Michaud et al., 2013). The extent to which infiltrating monocytes or monocyte-derived macrophages exist in brain parenchyma in AD and in models with AD pathology remains unclear. If infiltrating monocytes or monocyte-derived macrophages do enter the brain in AD, questions still remain to define their origin, how to distinguish them from brain resident activated microglia, and whether they are beneficial or detrimental to Aβ- or tau-related pathologies (El Khoury et al., 2007; Prokop et al., 2015; Silvin et al., 2022; Simard et al., 2006; Varvel et al., 2012; Wang et al., 2016).

Reconstructed adaptive niche in the brain parenchyma in AD

T and B cells express receptors with potential to recognize diverse antigens from pathogens, tumors, disease states, and also maintain immunological memory and self-tolerance (Akkaya et al., 2020; Kumar et al., 2018; Nemazee, 2017). Several studies have found an increase of T cells in the CSF, leptomeninges, and hippocampus in AD patient post-mortem tissue and in both Aβ and tau mouse models (Gate et al., 2020; Laurent et al., 2017; Merlini et al., 2018). A predominance of CD8⁺ rather than CD4⁺ T cells was noted and a greater number of T cells were located in hippocampus and other limbic structures with more severe pathology in AD patients, indicating a tight link between neuronal damage and T cell accumulation. However, neuronal loss also occurred in the hippocampus of non-AD dementias, and T cells were not consistently associated with plaques or NFTs (Togo et al., 2002). In another study, CD8⁺ T cells were observed in the brains of AD patients, especially in the hippocampus with tau but not Aβ pathology (Merlini et al., 2018). These findings indicate a potential role of T cells in AD progression, but the accumulation of T cells in brain parenchyma requires a complex disease and immune milieu. Interestingly, infiltration of T cells has also been observed in aged mice and humans. T cells in the ageing brain were clonally expanded and differed from those in corresponding peripheral blood, suggesting they may recognize specific antigens in the ageing brain (Dulken et al., 2019). Several reports have implicated an increase of T cells in the brain responding specifically to α-synuclein during the neurodegenerative process of Parkinson’s disease in mouse models and patients (Gate et al., 2021; Lindestam Arlehamn et al., 2020; Sulzer et al., 2017). This indicates a potentially shared adaptive immune responses in neurodegenerative diseases. Whether specific antigens, such as Aβ peptide, variously modified forms of tau, or protein or myelin debris released by damaged neurons are presented to adaptive immune cells within AD remains a major intriguing question. Sequencing TCR at the single cell level combined with high-throughput peptide screening would enable elucidation of the specific antigens, which might in turn yield pathological stage-specific therapeutic strategies.

Accumulating genetic and functional evidence strongly indicates both beneficial and detrimental roles of adaptive immunity in AD pathogenesis and disease progression, although it is unclear whether these effects are direct or indirect. Genetic ablation of peripheral immune cell populations in Rag^−/−−5xFAD mice, which lacked T cells, B cells, NK and NKT cells, significantly accelerated Aβ pathology, enhanced neuroinflammation, increased microglial number, decreased microglial process branching, and decreased microglial phagocytosis. In contrast, replacement of IgG or bone marrow transplantation reversed these effects and reduced Aβ pathology, suggesting the involvement of peripheral adaptive immunity in the pathogenesis of AD (Marsh et al., 2016). In another study, fibrillar Aβ deposits and total Aβ were significantly reduced in 8 month Rag2^−/−-APP/PS1ΔE9 mice, while microgliosis and Aβ phagocytosis were enhanced with Rag2^−/− bone barrow reconstitution of APP/PS1ΔE9 mice, suggesting decreased Aβ pathology with lifelong or acquired absence of T and B cells (Spani et al., 2015). Depletion of CD8⁺ T cells by anti-CD8 antibody treatment neither changed plaque pathology nor improve cognitive performance, but altered neuron- and synapse-related gene expression in 12 month old APP/PS1ΔE9 mice (Unger et al., 2020). Regulatory T cells (Tregs) play an essential role in suppressing excessive immune responses and self-tolerance (Sakaguchi et al., 2008). Elevated FoxP3⁺ Treg cells were observed in the periphery of 5xFAD mice and AD patients (Baruch et al., 2015; Rosenkranz et al., 2007). Depletion of Treg cells by crossing mice with FoxP3-diphtheria toxin (DTx) receptor followed by DTx administration or pharmacological inhibition of Tregs increased infiltration of monocytes/monocyte-derived macrophages and T cells, with mitigated Aβ pathology, astrocyte activation and improved spatial learning in 5xFAD mice (Baruch et al., 2015). Th1 and Th17 cells were present in the parenchyma of the APP/PS1 mice, and Aβ-specific Th1 rather than Th17 cells produced high levels of IFN-γ, which increased CD11b expression and Aβ deposition, and impaired cognitive function in these mice (Browne et al., 2013). The opposite results were observed in APP/PS1ΔE9 mice ICV-injected with Th1 cells, which reduced plaque load (Fisher et al., 2014). B cells are also a heterogeneous population. Although no direct evidence has been shown that B cells are present in the brain parenchyma in AD models, activated B cells, including CD5⁺CD11b⁺CD19⁺ B1a and CD5^-CD11b⁺CD19⁺ B1b cells were increased in the spleen and cervical lymph nodes in aged 3xTg AD mice. 3xTg or APP/PS1 mice were produced that were deficient in B cells by crossing with J_HT mice. The mice showed improvement on cognitive tests and locomotion activity. Aβ burden was reduced and TGFβ⁺ microglia increased in B cell deficient AD mice. Depletion of activated B cells by anti-CD20/B220 antibody ameliorated Aβ related AD pathology in 5xFAD mice (Kim et al., 2021).

Taken together, in addition to their direct effects on neuronal viability, adaptive immunity also affects pathology and interacts with innate immunity, which in turn can influence neurodegeneration, indicating adaptive immunity is an important component in Aβ pathogenesis. However, additional questions remain to be answered, some of which are related to 1) the temporal features of changes in adaptive immunity during AD progression and the appropriate time window to consider targeting the adaptive immune system for intervention; 2) specific lymphocyte subpopulations, including but not limited to, effector CD4⁺ T cells (Th1, Th2, Th17), Treg, cytotoxic CD8⁺ T cells, exhausted CD8⁺ T cells, brain resident T cells and other major potential immune mediators in AD; and 3) the cross talk between adaptive immunity and innate immunity in various stages of AD.

Interplay between innate and adaptive immunity in AD.

Under homeostatic conditions, Treg, and effector T cells interact with microglia together with their secreted anti- or pro- inflammatory cytokines, including IL-10, TGF-β, IL-4, and IFN-γ (Schwartz et al., 2013). Brain resident cells in the neurodegenerative or aged niche, such as neuron, microglia, astrocytes, endothelial cells or macrophages can produce cytokines, neurotoxic reactive oxygen species (ROS) and inducible nitric oxide synthase (iNOS), leading to a neuroinflammatory cascade in brain parenchyma, that can lead to T cell invasion (Mrdjen et al., 2018; Pober and Sessa, 2007; Salter and Stevens, 2017). Meanwhile, with a changing brain milieu including changes to brain barriers, the meninges (leptomeningeal vessels), blood-brain barrier (parenchymal vessels) and blood-CSF barrier (choroid plexus) can open the gate to T cells due to their higher expression of adhesion molecules ligands and integrins. Those activated T cells secrete pro-inflammatory and neurotoxic mediators to perpetuate the inflammatory cascade and induce microglial state transition, which may initiate or accelerate brain atrophy. We and others have found a population of microglia that is present particularly in the presence of tau pathology which have very high expression of CD80, CD86, MHC-I, and MHC-II complex genes, such as H2-D1, B2m and H2-DM (Mathys et al., 2017; Neefjes et al., 2011; Rexach et al., 2020). Whether this “dendritic cell-like” microglial population has a role in antigen presentation to T cells in such an AD-like disease state and their interaction with subsets of T cells require further studies (Figure 2).

The regional progression of brain atrophy in AD highly correlates with tau accumulation but not amyloid deposition (Giannakopoulos et al., 2003). Whether T cells are related to or directly contribute to neurodegeneration remains unknown in AD in model systems with AD type pathologies. If they turn out to be involved in mediating neurodegeneration, whether they could serve as modulatory targets for treating AD is also an intriguing question. Immune checkpoint blockade targeting the PD-1/PD-L1 pathway leads to modification of common factors that go awry in AD and dementia. One study suggests that this pathway may be involved in cognitive function in both Aβ and tauopathy models (Rosenzweig et al., 2019). T cells are the major population of cells releasing IFNs, especially IFN-γ, and thus after their local infiltration and cytokine release, the resident innate immunity and adaptive immunity might have very strong intercellular communication and therefore play important roles in disease states. Whether interferons and specifically IFN-γ signaling could serve as a potential therapeutic target by either neutralization of IFN-γ or genetic or other manipulation of IFNGR, the IFN-γ receptor in specific cell types, such as microglia and neurons, require further studies. Hence, the role of adaptive immunity with tauopathy and neurodegeneration together with the immune microenvironment in the brain parenchyma should be further explored in future studies. This would allow for detecting direct evidence for whether or not adaptive immunity is influencing neurodegeneration.

Cross talk between the periphery and the brain

Obviously, the brain does not function in isolation, and the periphery and brain may cross talk in both directions. The cellular and molecular basis involving how communication occurs between the periphery and the brain might include but are not limited to vagal-nerve sensory afferents, brain originated substances entering the blood and communicating with peripheral immune cells, peripheral immune cells or active factors communicating with brain endothelium or vascular macrophages or microglia, and direct infiltration of peripheral immune cells into the brain parenchyma (Chakravarty and Herkenham, 2005; Laflamme and Rivest, 1999; Perry et al., 2007; Tracey, 2002). Of note, studies have identified that ageing, diabetes, circadian rhythm and sleep dysfunction as well as gut microbiota are AD risk factors, which affect systemic infections and inflammation and the progression of chronic neurodegeneration, including AD. Peripheral inflammatory stimuli induced immune training or tolerance in microglia, in turn promoted or alleviated Aβ deposition separately by injecting either single or four consecutive days of LPS to 3 month old APP23 mice and analyzed after 6 months. As a result, epigenetic modifications in microglia led to upregulated HIF-1 and mTOR signaling in response to peripheral immune training stimulation (Wendeln et al., 2018). Systemic inflammation altered microglial morphology and impaired Aβ clearance in an NLRP3-dependent manner in APP/PS1 mice (Tejera et al., 2019). Increased wakefulness also caused increased neuronal activity and elevated Aβ production with suppressed glymphatic system function and Aβ and tau clearance (Musiek and Holtzman, 2016). Gut microbiota regulated Aβ pathology and microglial morphology. Early lifetime antibiotic perturbations of the gut microbiome in male APPPS1-21 mice led to reductions in Aβ plaque pathology and altered phenotypes of plaque-associated microglia (Dodiya et al., 2022).

Concluding remarks and prospective directions

The complex nature of the CNS necessitates its own specialized immunological adaptations to detect and respond to environmental changes. Under homeostatic conditions, microglia, and other innate immune cells, reside in and surveil the brain parenchyma to remove cell debris and maintain the local microenvironment. This is likely crucial for normal synaptic activity and connectivity. An intact BBB, meningeal layers, CSF/ISF fluid flow, and the dural lymphatic system work together to safeguard the brain as relatively “immune-privileged” and to serve as a border immune cell reservoir to communicate with the immune system in the periphery and the brain parenchyma. In AD, accumulation of aggregated forms of Aβ and tau as well as damage induced by aggregates of these proteins sequentially induce innate immune system activation in the brain parenchyma that includes reactive changes to astrocytes and microglia, changes to immune cells at the brain’s borders including BBB pathology, meningeal immune cell changes, and lymphatic drainage system dysfunction. Meanwhile, dysfunction of these safeguard structures might lead to ectopic infiltration of peripheral immune cells, especially cells of the adaptive immune system into the brain parenchyma and lead to an entirely new immune milieu in AD. There are changes of both the innate and adaptive immune systems in AD that are present with different pathologies, disease phases, and severities in which both anti-inflammatory and pro-inflammatory effects are occurring and likely contributing in important ways to AD development and progression. Concerted activities of both innate and adaptive immune effectors are crucial for shaping an appropriate immune response. A better understanding of the roles for adaptive immunity in the landscape of CNS immune milieu will deepen our understanding related to how the adaptive immune effectors and the interplay of the innate and adaptive immune cells contribute to AD development and progression.

Despite substantial advances that have been made in our understanding of AD pathophysiology and development of potential therapeutic interventions particularly targeting Aβ, the fundamental mechanisms underlying the clinically symptomatic phase of AD are still elusive. A better understanding of the specific roles of both the innate and adaptive immune system throughout the course of AD pathology, especially during the tau phase of AD, will likely lead to novel therapeutic intervention strategies for both the preclinical as well as clinically symptomatic phase of AD. Mapping the disease state-specific interlink between innate and adaptive immunity, especially microglia and T cells, including how they communicate, present antigens, and their pathophysiological responses will be a key nexus to set up unique therapeutic interventions to prevent or reverse neurodegeneration in AD and primary tauopathies.