The Role of Innate Immunity in Alzheimer’s Disease

AD pathology:

Alzheimer’s disease (AD) is a neurodegenerative disease characterized by neuronal loss, neuroinflammation, and pronounced memory decline.^1,2 Risk factors for developing this neurodegenerative disease include age, genetics, sex, brain injury, and various environmental factors.³ Regardless of origin, neuropathologies consistent among AD patients are amyloid beta (Aβ) and neurofibrillary tangle (NFT) deposition.⁴ Aβ can deposit in the brain decades prior to clinical symptom onset and stimulate other AD pathologies such as hyperphosphorylated tau (p-Tau) aggregation which triggers subsequent NFTs and neuronal death.^5–9 These NFTs can mediate neuronal damage that propagates mild cognitive impairment (MCI) and dementia.¹⁰ Given that Aβ accumulation precedes NFT formation and memory decline by many years, the focus on Aβ production has dominated AD research.¹⁰ Aβ is a cleavage product of amyloid precursor protein (APP).¹¹ APP cleavage can follow two pathways: amyloidogenic and nonamyloidogenic. The former has a cleavage event spurred by β-secretase and γ-secretase to produce neurotoxic Aβ, while the latter undergoes APP cleavage by α-secretase and γ-secretase to generate distinct products, some of which have important neuroprotective functions.^11,12

The Aβ peptide can act as a building block to produce several distinct higher-order forms of Aβ, some of which are more neurotoxic than others.¹¹ For instance, monomers of Aβ peptide can accumulate to form oligomers that can travel throughout the brain and constitute the soluble fraction of Aβ levels in the AD brain.¹¹ Aβ oligomers are the most toxic form of Aβ and have been shown to incite synaptic dysfunction and neuronal cell death signaling.¹³ Aβ oligomers can seed Aβ fibrils to promote the formation of Aβ aggregates known as “plaques”.¹⁴ Monomers of the Aβ peptide can have varying lengths depending on their cleavage.¹¹ However, Aβ42 and Aβ40 are the most prevalent in AD pathogenesis, with Aβ42 being more likely to incorporate into Aβ aggregates.¹⁵ AD brains are also marked by dystrophic neurites.¹⁶ Dystrophic neurites exhibit axonal transport defects and are often found surrounding Aβ plaques.⁴ Recent evidence suggests that dystrophic neurites can directly contribute to AD pathology propagation by promoting β-secretase-mediated cleavage of APP and subsequent Aβ peptide generation.¹⁷

Biomarkers:

The identification of biomarkers relating to AD risk and pathology prior to clinical onset has become an important area of focus in hopes of manipulating AD onset and progression. For decades, the diagnosis of AD relied on post-mortem brain histological assessment to identify Aβ and NFT pathologies. However, recent technological advancements in genetic testing, PET imaging, and sampling of the cerebrospinal fluid (CSF) and blood have opened doors to the possibility of earlier AD diagnosis. There is great hope that the identification of reliable AD biomarkers will move the field toward preventative medicine; however, continued progress on this front is needed to make this a reality.

There are two classifications of AD based on disease origin and onset: (1) early-onset AD (EOAD) (also known as familial AD or FAD) and (2) late-onset AD (LOAD). EOAD accounts for 5% of all AD cases and is primarily caused by mutations in APP, presenilin-1 (PSEN1), and presenilin-2 (PSEN2), all of which can lead to elevated levels of total Aβ and increased production of Aβ40 and Aβ42.^18,19 EOAD often occurs in individuals younger than 65 years of age.²⁰ In contrast, the origin of LOAD is linked to an interplay between genetic mutations, environment, and neuroinflammatory responses.²¹ Interestingly, many of the mutations linked to LOAD risk are known to affect innate immune signaling pathways (Table 1). The most common variant associated with LOAD risk is apolipoprotein E gene (APOE) ε4 allele.¹⁸ APOE has three isoforms which include ε2, ε3, and ε4, with ε4 being the allele exerting the greatest AD risk.²² In homeostasis, APOE is synthesized for lipid transport.²³ However, during stress, APOE ε4 is markedly more susceptible to proteolytic cleavage to generate products that promote NFT formation and mitochondrial dysfunction, both of which can contribute to AD progression.²³ In addition, APOE has been reported to promote Aβ aggregation and is often found in large quantities within Aβ plaques, further implying its role in AD pathogenesis.^24,25 Patients with the APOE ε4 allele are at 3-times higher risk for developing AD.²² The combination of APOE ε4 and chronic inflammation further increases risk for these carriers, emphasizing an important link between genetics and inflammation in AD risk.²⁶ LOAD is also associated with mutations in receptors expressed by microglia, the brain’s resident immune cells.²⁷ Recent research shows that multiple microglial receptors are critically involved in Aβ clearance and that mutations in multiple microglial immune receptors dramatically increases the risk for LOAD.²⁷

Diagnosing AD in patients has become amenable with the use of positron emission tomography (PET) imaging of Aβ and tau in the brain.²⁸ PET imaging is able to detect high levels of Aβ in the brains of patients with cognitive impairment, allowing for highly sensitive diagnosis of AD.^29–31 In addition, the detection of tau using PET has also proven to aid in the tracking of AD progression.^28,32 An early diagnosis of AD is also made possible with PET imaging of the initial inflammation of disease, which is often characterized by microglial activation.³³ Biomarkers in the CSF and blood are also used to diagnose and detect AD progression.³⁴ CSF markers include levels of Aβ and tau, with AD patients presenting with a decrease in Aβ42 and a significant increase in phosphorylated and total tau.^35,36 Blood is an alternative medium to measure for AD biomarkers, and is especially attractive due to its low invasiveness and medical cost.³⁷ Blood biomarkers for AD include decreased levels of Aβ42, a lower Aβ42:Aβ40 ratio, and increased tau.³⁷ In addition, the level of neurofilament light chains (NfLs) is another blood marker for AD.³⁸ Increased levels of NfLs in the blood is distinct in EOAD patients prior to symptom onset, and changes in NfL levels are often greatest when AD patients transition from the presymptomatic to symptomatic stages of AD.³⁸ Therefore, NfLs can be used to predict AD onset and track its progression in a non-invasive manner.

Innate immune function of AD pathologies:

While numerous physiological functions have been assigned to Aβ and tau, there is still a general lack of consensus as to what roles these AD-related molecules play in homeostasis. Interestingly, it has recently been proposed that Aβ may help provide protection against pathogens by functioning as an antimicrobial peptide (AMP) (Fig. 1).³⁹ Such studies demonstrate that Aβ oligomers are critically involved in limiting both fungal and bacterial infections in cell culture, C. Elegans, and mice.⁴⁰ Additionally, bacterial infections have been shown to incite the seeding of Aβ in the 5xFAD AD mouse model.⁴⁰ Aβ’s AMP function may explain why the peptide has been conserved across a breadth of species for hundreds of millions of years.⁴¹ Further implicating Aβ in the innate immune response is the finding that certain microorganisms’ cell membranes are covered with amyloid-like fibrillous structures.⁴² Bacteria with these amyloid structures, also known as curli fibrils, present a pattern recognized by receptors on immune cells to initiate signaling to remove these invaders.^43,44 A similar response is elicited when Aβ itself is recognized by microglial pattern recognition receptors (PRRs) in the brain.⁴⁵ Unfortunately, chronic innate immune activation augmented by sustained Aβ recognition provokes subsequent inflammation that can lead to further propagation of AD pathologies and neurodegeneration.^2,46

Although the explicit link between Aβ-seeding and tau accumulation has remained elusive, recent research has begun to shed light on this topic.⁵ For instance, it has been demonstrated that Aβ-induced dystrophic neurites accumulate endogenous tau that eventually form into neurofibrillary tangles that correlate with cognitive deficits.⁵ In homeostasis, endogenous tau participates in the assemblage and stabilization of microtubules in neurons, and, as a result, tau is believed to play functionally important roles in neuronal transport.⁴⁷ However, the detachment of tau from microtubules and hyperphosphorylation of tau can incite aberrant microglial activation and proinflammatory cytokine production.⁴⁸ This feedback drives further production of pathogenic tau to enter a cycle that perpetuates its overproduction and neurotoxicity.⁴⁸ Interestingly, herpes simplex virus (HSV-1), which infects two-thirds of the global population, has been shown to phosphorylate tau.^49,50 In accordance with this finding, antiviral medication has been reported to decrease phosphorylated tau accumulation and Aβ load.⁵¹ A clinical trial is currently investigating the efficacy of the antiviral drug valacyclovir for AD patients with HSV-1 or HSV-2.⁵² This further suggests a link between both neuropathologies of AD and the innate immune response that requires further exploration.

Microglia receptors implicated in AD:

Microglial clearance of Aβ and tau:

Clearance of AD pathology by microglia is executed by several discrete processes. These include receptor-mediated phagocytosis and endocytosis, expression of degradative factors, extracellular chaperone-mediated clearance, and induction of autophagy.¹¹¹ Each mechanism allows for homeostatic maintenance of the brain parenchyma, and can potentially be targeted to ameliorate Aβ and tau accumulation in disease.

Disease associated microglia:

The dispute surrounding whether microglia exert beneficial or detrimental roles in AD remains a hotly debated topic. Single-cell RNA-seq recently identified a new subset of microglia known as “disease associated microglia” (DAM).⁶³ In the past, microglia were largely thought of as a homogenous cell population. However, this diversified transcriptional profile of microglia in the AD and aging brain environment has prompted many new questions surrounding their role in disease onset and progression. DAMs are classified by two distinct stages: the initial TREM2-independent Stage 1 and TREM2-dependent Stage 2.⁶³ Stage 1 and 2 DAMs are identified by their downregulation of characteristic homeostatic microglial genes (i.e. P2ry12, Cx3cr1, and Tmem119). Stage 1 DAMs upregulate phagocytic related genes such as Apoe and Dap12 while Stage 2 DAMs upregulate Clec7A and Trem2.⁶³ As previously discussed in this review, APOE variants are a major risk factor for LOAD, while TREM2, DAP12, and CLEC7A are receptors previously described in phagocytic pathways.^23,65,87 Initiation of this shift of microglia from a homeostatic phenotype to the DAM signature is believed to be dependent on sensing of NAMPs, engagement of TREM2 signaling, and extended interactions of microglia with Aβ plaques.⁶⁴

The DAM subtype is accepted to serve in a beneficial capacity through their coordination of Aβ clearance.⁶³ However, this function remains strictly regulated by the inhibitory signaling of CX3CR1, perhaps acting as a harness on DAM function.⁶³ Therefore, future research efforts will likely focus on stimulating the DAM phenotype early on in disease to reduce AD progression. However, the possibility also exists that over-excessive and/or chronic activation of the DAM phenotype may trigger neurotoxic consequences and propagate AD pathology. This possibility will likely encourage a new AD-microglia field focused on determining how to revert DAM back to their homeostatic microglial state. The multi-faceted roles of microglia underscore the temporal complexities of their function, and will likely keep them in the AD research spotlight for years to come.

Loss of microglial homeostatic function in AD:

Homeostatic microglia possess a signature defined by the expression of receptors such as CX3 chemokine receptor 1 (CX3CR1) and purinergic receptor P2RY12.²⁷ CX3CR1, which is activated in response to its ligand fractalkine, performs inhibitory duties to sustain a resting-state microglia population.¹³⁷ Although fractalkine is found at higher levels in the blood of AD patients, Cx3cr1 is downregulated in DAMs.^63,138 CX3CR1 is thought to have complex roles in AD. For example, deficiency in Cx3cr1 reduces neuronal loss in the 3xTgAD mouse model, which develop both Aβ plaques and NFTs.¹³⁹ In contrast, the loss of Cx3cr1 in the Aβ-mediated hAPP-J20 AD mouse model spurs microglial toxicity and intensifies cognitive deficits.¹⁴⁰ The presence of Aβ also downregulates fractalkine in the hAPP-J20 AD mouse model.¹⁴⁰ These findings suggest that the role of CX3CR1 is variable depending on the different models of AD being used. Despite these inconsistencies, CX3CR1 appears to be a critical modulator of DAM biology. However, without this CXC3R1 check and balance, excessive microglia activation may contribute to neuronal loss and functional decline, while its overabundance may act to inhibit necessary DAM response to AD pathology. Purinergic receptor P2RY12, on the other hand, is important for microglial migration, and it is highly expressed in homeostatic microglia.^141,142 However, the role of P2RY12 downregulation in plaque-associated DAMs and AD remains unclear and requires further investigation.

Impact of microglial elimination in AD:

The often-contradictory roles of microglia in varying states and stages of neurodegenerative disease have prompted researchers to explore the impact of microglia depletion in AD. In recent years, treatment with the colony stimulating factor 1 receptor (CSF-1R) inhibitor, Plexxikon, has emerged as a popular approach to study the functional role of microglia in vivo. CSF-1R signaling is required for the survival of microglia and other peripheral macrophages, and as a result, Plexxikon treatment results in rapid and robust depletion of microglia from the brain.¹⁴³ Results from recent studies using Plexxikon indicate a complex role for microglia in mouse models of AD that is greatly influenced by the timing and efficacy of depletion as well as the AD mouse model studied. For example, when 3xTgAD mice were depleted of 30% of their microglia, and aged to have existing Aβ plaques and tau aggregation, the mice were shown to have improved cognition but this partial deletion of microglia had no effect on Aβ plaque load or tau aggregation.¹⁴⁴ In comparison, a 90% microglial depletion in PS19 AD mice, aged to have tau aggregation, led to significant reductions in tau propagation.¹⁴⁵ Meanwhile, partial depletion of microglia in the APPPS1 mouse model of AD was found to improve memory and ameliorate synaptic degeneration, while having no effect on Aβ load.¹⁴⁶ 5xFAD mice depleted of 80% of their microglia similarly showed improved memory and decreased neuronal damage without modulation of Aβ levels.¹⁴⁷ However, 5xFAD mice depleted of greater than 97% of their microglia prior to Aβ plaque seeding-onset (i.e. Plexxikon treatment beginning at 1.5 months of age) exhibited reductions in Aβ load as well as a rescue of anxiety-related behaviors.¹⁴⁸ However, the almost complete depletion of microglia in these studies was also found to promote increased numbers of dystrophic neurites, as microglia were no longer present to protect surrounding neurons from the neurotoxicity of the few remaining Aβ plaques.¹⁴⁸ Taken together, current data in mice indicates that partial depletion of microglia does not affect Aβ plaque and tau load but can result in improved memory and less neuronal damage compared to controls.^144,146,147 However, near-complete depletion of microglia amid or prior to AD pathology onset attenuates Aβ and tau load.^145,148 While these studies all point to critical roles of microglia in AD, they also highlight that microglia-targeting therapeutics must carefully take into account the status of disease and desired microglia depletion efficacy to maximize beneficial clinical outcomes and safety.

AD immune cells: microglia or macrophages?

These intriguing findings from Plexxikon depletion studies have prompted speculation over whether microglia or peripheral monocytes repopulate this niche when the depletion drug is removed. As mentioned above, CSF-1R inhibition with Plexxikon treatment can deplete microglia, peripheral macrophages in other tissues, and potentially other cell types reliant on CSF-1R signaling for survival.¹⁴³ This has led to speculation over the discrete roles of microglia versus other peripherally derived cell types in AD. Plexxikon microglial depletion treatment by itself does not compromise the BBB, and expansion of the few remaining microglia is thought to repopulate the brain following cessation of Plexxikon treatment.^149,150 However, injury models, such as traumatic brain injury (TBI), can cause BBB breakdown and subsequent infiltration of immune cells into the brain.^151,152 Likewise, AD pathology is also thought to propagate the loss of BBB integrity.¹⁵³ Microglial depletion and repopulation after TBI has been reported to replace chronically activated microglia with resting-state myeloid cells, resulting in decreased neuroinflammation and neuronal loss.¹⁵¹ The neuroprotection conferred by the repopulated myeloid cells post-TBI relies on their induction of IL-6 trans-signaling to enhance learning and memory.¹⁵⁴ Therefore, the replacement of activated microglia with more quiescent myeloid cells may underlie neuroprotection. However, while it is speculated that depleted microglia die and are cleared away following Plexxikon treatment, the identity of the repopulating cells in BBB-compromised settings remains poorly understood.¹⁴⁹ Multiple myeloid cell types have been proposed to fill this niche including peripheral macrophages and microglia from the small Plexxikon-resistant subset. However, future studies are needed to better understand this process and how it affects neurological disease. This further illuminates the complexity and heterogeneity of CNS macrophages, and maintains speculation surrounding their capacity to be manipulated during neurodegenerative disease.

Initial studies exploring the ability of CNS-infiltrating monocytes to influence AD-related disease in the APPPS1 AD mouse model initially showed that chemokine receptor CCR2 deficiency prevents monocytes from entering the brain and that this results in fewer IBA-1⁺ myeloid cells in the brain and impaired Aβ clearance.¹⁵⁵ These data suggest that peripheral monocytes are important for properly controlling AD pathology. In contrast, CCR2 deficiency in the same mouse model was later shown by a different lab to increase plaque-associated microglia numbers and have minimal impact on Aβ accumulation.¹⁵⁶ What underlies the discrepancies between these two studies remains to be fully resolved. One possibility explaining the divergent outcomes may be related to the latter study’s use of irradiation, which can encourage the recruitment and retention of peripherally-derived monocytes in the brain.¹⁵⁷ Notably, recent research utilizing an irradiation-independent model has established brain engrafting macrophages that maintain a distinct transcriptional profile from microglia and are able to repopulate the brain following partial microglia depletion.¹⁵⁸ Other work also demonstrates that TREM2 can promote peripheral monocytes to enter the brain in APPPS1 mice.¹⁵⁹ In contrast to the belief that peripheral myeloid cells are critical for proper AD control, parabiosis studies with WT and 5xFAD mice show minimal infiltration of peripheral monocytes from WT mice into 5xFAD brains.⁷¹ Likewise, these parabiosis findings have also been confirmed in the APP/PS1 mouse model.⁷¹ In addition, recent research using tdTomato lineage tracing in 5xFAD mice convincingly shows that CX3CR1⁺ microglia comprise virtually all of the myeloid cells surrounding the Aβ plaques, with virtually no CCR2⁺ peripheral monocytes found in the vicinity of Aβ in the brain.⁵⁷ In terms of Aβ aggregation mouse models, which include 5xFAD and APPPS1 transgenic mice, microglia appear to be the key myeloid cell type responding to Aβ, with little assistance from infiltrating monocytes. However, this work cannot exclude the potential of peripheral monocyte assistance under all conditions associated with human neurodegenerative disease, especially those that are not fully recapitulated in mouse models. In particular, it will be important for future studies exploring the roles of discrete myeloid cell lineages in AD to assess how combined Aβ and tau pathology can influence outcomes.

Cytokines in AD:

Multiple innate immune cytokines including type I interferon (IFN), TNF-α, IL-18, IL-1β, IL-33, IL-34, IL-12, IL-23, and IL-10 have been described to play important roles in various aspects of AD pathogenesis.^163–170 Resting levels of pro-inflammatory cytokines are found to be higher in the blood and cerebrospinal fluid (CSF) of AD patients, suggesting a role for them in disease pathogenesis.¹⁷¹ The deposition of Aβ, and resultant neuronal damage, provide ligands to activate microglial phagocytosis of Aβ while also promoting microglia-mediated production of proinflammatory cytokines.⁵⁸ Acutely, this process can lead to the proper clearance of Aβ allowing the brain parenchyma to restore homeostasis (Fig. 3).¹⁷² However, these inflammatory mediators can have complex effects in AD. For example, under chronic inflammation, this process can lose its effectiveness in clearing amyloid beta, while continuing its inflammatory response to further propagate neurotoxicity in a feed-forward mechanism (Fig. 3).¹⁷³ Catalysts for this continued and detrimental response include factors such as age, genetics, injury, and peripheral inflammation.² The production of specific pro-inflammatory cytokines, such as IFN-γ, also has the potential to encourage further production of Aβ as well as to recruit microglia activation to respond to neurotoxic Aβ species.¹⁷⁴ Therefore, cytokine production can promote a cycle that is difficult to resolve.

As was alluded to above, IFN cytokines play significant roles in shaping microglial responses in AD and aging.^163,175 Several AD mouse models show a distinct IFN-stimulated gene (ISG) signature.¹⁷⁶ IFN has been described to be expressed by immune cells sensing nucleic acids.¹⁷⁷ Of note, fibrillar amyloid beta often contains nucleic acids that can elicit an inflammatory ISG signature in microglia.¹⁷⁶ This neurotoxic type of microgliosis has been shown to promote complement-mediated synapse loss that can be curtailed with IFN-receptor blockage.¹⁷⁶ Researchers have also shown that knocking out IFN-γ receptor type I in transgenic APP Swedish mutant AD mice leads to a reduction in Aβ plaque load and reactive microglia.¹⁷⁴ In addition, in vitro studies indicate that IFN-γ promotes microglial release of TNF-α, and both of these cytokines promote β-secretase cleavage of APP to produce more Aβ.¹⁷⁴ These results strongly suggest that IFN is associated with AD propagation.

The role of TNF-α in AD has remained enigmatic, as some research has focused on the therapeutic benefit of its inhibition, while others show beneficial or benign roles for the cytokine. In support of a detrimental role for TNF-α in AD, it has been shown that higher CSF levels of TNF-α correlate with impaired “functional connectivity,” a measure of the communication between different brain regions underlying various functions such as learning, memory, inhibition, etc.¹⁷⁸ In line with pathologic functions of TNF-α in AD, APPPS1 mice deficient in TNF receptor 1 (TNFR1) have been described to have decreased levels of Aβ, lower expression of inflammatory factors, improved maintenance of choroid plexus tissue and CSF-blood barrier integrity, and preserved memory compared to TNFR1 sufficient APPPS1 mice.¹⁷⁹ In contrast, an injection of murine TNF-α AAV into the hippocampus of APP transgenic mice has been reported to attenuate Aβ deposition and incite increased microglial responses without increasing APP levels.¹⁸⁰ This suggests that under certain spatial and temporal contexts TNF-α may promote plaque clearance. Finally, the proposed causal relationship between TNF-α and AD has recently come into question, as Mendelian randomization modeling of GWAS studies show no causality between peripheral TNF-α expression levels and the risk of developing AD.¹⁸¹

IL-1 family cytokines, such as pro-inflammatory IL-18, IL-1β, and IL-33, have also been implicated in modulating AD pathogenesis.^167,182,183 For example, IL-18 promotes β-secretase mediated cleavage of APP to produce more Aβ40.¹⁸⁴ However, loss of IL-18 in the APPPS1 AD mouse model was found to incite seizure through a proposed mechanism that involves excessive excitatory synapse activity and impaired dendritic pruning.¹⁸⁵ IL-1β has conflicting roles in AD pathologies. Increased peripheral levels of IL-1β and its propensity to incite neuroinflammation have led scientists to focus on neutralizing its functions in hopes of therapeutic benefit in AD.^171,186 For instance, IL-1 receptor inhibition in 3xTgAD mice was reported to rescue cognitive deficits, decrease tau phosphorylation, and modestly modulate Aβ levels.¹⁸⁶ Interestingly, conditional overexpression of IL-1β in 3xTgAD mice was conversely found to markedly decrease Aβ levels while concomitantly leading to increased tau phosphorylation, further highlighting the complex role of neuroinflammation in AD.¹⁸⁷ Increased IL-1 production is also known to promote S100B protein release from astrocytes.¹⁸⁸ The S100B protein is a neurotrophic cytokine that often acts as a pro-inflammatory cytokine in AD.¹⁸⁹ Higher levels of S100B have been documented in AD patients and appears to be associated with increased Aβ plaque load and gliosis.^190–192

IL-33, an alarmin in the IL-1 cytokine family, has also been implicated in AD. IL-33 is expressed by oligodendrocytes and its receptor, ST2, is often expressed by microglia.¹⁹³ IL-33 expression is reduced in the brains of AD patients, identifying this cytokine as a potential modulator of disease.¹⁹⁴ Accordingly, intraperitoneal injection of IL-33 was recently shown to restore LTP deficiency, improve contextual memory, and decrease Aβ levels in APPPS1 AD mice.¹⁶⁷ IL-33 treatment accomplishes this rescue by increasing microglial Aβ phagocytosis and anti-inflammatory gene expression, while also decreasing microglial pro-inflammatory gene expression.¹⁶⁷

IL-34 and CSF-1 activate the CSF-1 receptor (CSF-1R) to promote microglial population survival and maintenance.¹⁴⁹ Interestingly, AD patients have been shown to have increased expression of IL-34 in the white matter compared with age-matched controls, highlighting its importance in regulating AD pathogenesis.¹⁹⁵ More recent work, however, shows reduced expression of IL-34 in the inferior temporal gyrus (ITG) and an increase in expression of CSF-1 in the ITG and middle temporal gyrus (MTG) of AD patients.¹⁹⁶ On one hand, research has found neuroprotective roles for IL-34 in AD. For instance, intracerebroventricular injection of IL-34 into APPPS1 mice promotes improved memory and lower levels of oligomeric Aβ through a mechanism that involves increased expression of insulin degrading enzyme and heme-oxygenase-1.¹⁶⁸ In contrast, using microglial cultures from AD human brains, researchers show that IL-34 stimulation decreases expression of genes associated with phagocytosis such as CD68, which is a lysosomal marker commonly used to evaluate phagocytic potential in AD mouse models.¹⁹⁶ Collectively, these findings highlight the potential temporal and anatomical complexity of IL-34 in the AD brain.

Increasing evidence also supports the role of IL-10, IL-12, and IL-23 in AD pathogenesis. For example, the shared subunit of IL-12 and IL-23 (i.e. p40) and IL-10 are found at elevated levels in the CSF of AD patients.^197,198 A blood biomarker study also found IL-10 and p40 at increased levels in cognitively normal patients that exhibit abnormally high levels of Aβ deposition.¹⁹⁹ In turn, the ablation of p40 can attenuate cognitive deficits and Aβ levels in APPPS1 mice.¹⁹⁷ Peripheral treatment with anti-p40 antibody in APPPS1 mice is also able to decrease Aβ levels.¹⁹⁷ Interestingly, the loss of IL-12 and IL-23 signaling through p40 genetic ablation has sex-specific effects in which only males exhibit a reduction in Aβ plaques.²⁰⁰ The silencing of p40 in aged senescence-accelerated mouse prone-8 (SAMP8) mice also mitigates cognitive decline, Aβ load, and neuronal death.¹⁶⁹ Lastly, ablation of IL-10 in the context of AD is also considered beneficial. In fact, APPPS1 mice lacking IL-10 show improved synaptic health and cognition compared with APPPS1 controls.¹⁷⁰ Therefore, the unchecked functioning of these cytokines during chronic inflammation of AD poses a serious threat to brain health and function.

Role of inflammasomes in AD:

Recent studies suggest a detrimental role for inflammasomes in AD.^201–204 Inflammasomes are multimolecular complexes that form following receptor activation in response to various DAMPs, PAMPs, and NAMPs.²⁰⁵ The formation of these complexes initiates a signaling cascade in which active caspase-1 promotes the cleavage of inflammatory pro-IL-1β and pro-IL-18 cytokines into their active forms and also incites cell death.²⁰⁵ Inflammasome-mediated cell death is termed “pyroptosis” and is executed by cleaved gasdermin-D.²⁰⁶ Gasdermin-D proteins are cleaved by caspase-1,−4,−5, or −11, and this product comes together to form a pore on the cell membrane to promote cell lysis and the exit of pro-inflammatory cytokines from the cell.²⁰⁷ The NLRP3 inflammasome, in particular, has been identified as an important contributor to AD pathogenesis.^{185,201,202,208–210} NLRP3 is upregulated and activated in AD, with increased mRNA and protein expression in immune cells.²¹¹ The loss of NLRP3 inflammasome in APPPS1 mice attenuates spatial memory and LTP deficits, in addition to decreasing Aβ burden.²⁰¹ NLRP3 was also recently shown to promote tau hyperphosphorylation and aggregation.^212,213 Accordingly, inhibitors such as the JC-124 drug or fenamate class NSAIDs have been developed to successfully inhibit the NLRP3 inflammasome to mirror these improvements in a potentially therapeutic format.^202,214 Recent research shows that apoptosis-associated speck-like protein containing CARD (ASC), an adaptor protein critical for the formation of the NLRP3 inflammasome, binds with Aβ.²¹⁵ This ASC-Aβ interaction initiates NLRP3 inflammasome activation and prevents proper clearance of Aβ.²¹⁶ In turn, the increased Aβ levels promote microglial pyroptosis leading to the extracellular release of more ASC that can bind Aβ to activate the cycle again, generating a chronic neuroinflammatory cascade.²¹⁶

The AIM2 inflammasome, which is activated by double-stranded DNA, has also been implicated in AD.²⁰⁴ Aim2 deficient 5xFAD AD mice have decreased Aβ plaque load, however, this is not sufficient to rescue memory deficits. Additionally, 5xFAD;Aim2^−/− mice have increased inflammation mediated by preserved IL-1 expression and an increase in the expression of pro-inflammatory cytokines IL-6 and IL-18 in the brain.²⁰⁴ This may suggest parallel inflammatory mechanisms to compensate for the loss of Aim2 in AD mouse models. Interestingly, our lab and others have shown the loss of Aim2 impacts neuronal morphology and negatively influences several behaviors such as anxiety and memory.^217,218 Furthermore, our recent studies demonstrate that defects in the AIM2 inflammasome can alter neurodevelopment.²¹⁷ Thus, it is feasible that changes in brain development existing before Aβ deposition may influence disease outcomes in AD mouse models. Moving forward, it will be interesting to revisit the role of the AIM2 inflammasome in AD using approaches that conditionally ablate AIM2 signaling following neurodevelopment.

Role of age, environmental factors, and sex in AD:

In recent years, there has been increasing appreciation for the instrumental roles that aging, environmental factors, and lifestyle choices play in AD pathogenesis (Fig. 4).²¹⁹ Interestingly, emerging work has shown that many of these environmental and lifestyle factors influence AD progression by modulating aspects of inflammatory signaling.² In this section, we will spotlight up-and-coming work describing how changes to inflammatory responses are now believed to underlie the ability of many environmental and lifestyle factors to influence AD.

Age is by far the greatest risk factor for developing AD.²²⁰ Interestingly, recent studies suggest that aging profoundly impacts microglial biology and that this can hinder their ability to properly handle and dispose of Aβ.⁸⁶ Along with their propensity to take on the DAM transcriptional profile, aging also alters microglia morphology, enhances their activation status, limits their motility, and curbs their ability to phagocytose Aβ.^63,221–223 Unfortunately, all of these age-related microglial changes have the potential to promote AD pathology. The ability of microglia to self-sustain their population is also speculated to make them particularly susceptible to telomeric shortening and cellular senescence, and it has been proposed that the acquisition of this senescence phenotype can cause microglial dysfunction.²²⁴ An additional mechanism by which aging contributes to AD is the loss of “youthful” blood-derived factors as one ages. Such studies demonstrate that parabiosis or plasma transfer from young mice to old mice improves learning and memory.^225,226 Two factors shown to be elevated in young blood include thrombospondin-4 and SPARCL1, which support synaptic connectivity.²²⁷ It has also been shown that blood from old mice negatively impacts cognition in young mice.²²⁸ One factor suggested to contribute to this phenomenon is the chemokine CCL1, as it is elevated in aged-mice plasma levels and directly impairs cognition in recipient young mice.²²⁸

Obesity and exercise have also been implicated in AD prevalence and severity.^{166,229–232} Researchers have shown that obese individuals have high levels of the short chain fatty acid known as palmitate in the CSF and that this can lead to impaired neuronal insulin signaling.¹⁶⁶ They further report that increased levels of palmitate in the CSF of aged mice provokes pro-inflammatory TNF-α cytokine production and that this can lead to memory deficits.¹⁶⁶ Indeed, they showed that TNF-α receptor deficient mice treated with palmitate perform better on memory tests compared with palmitate treated wildtype controls.¹⁶⁶ In addition, an intracerebroventricular injection of infliximab, a TNF-α neutralizing antibody, was also found to improve memory and ameliorate microgliosis in mice on a high fat diet.¹⁶⁶

Routine exercise, on the other hand, has been shown to help limit AD development and progression²³³. One mechanism that may explain this phenomenon is the release of the myokine FNDC5/irisin from the muscle to the brain during exercise.²³⁰ FNDC5/irisin production following exercise increases synaptic plasticity, a critical aspect of memory formation and maintenance^230,234. A mechanism by which memory formation is promoted by irisin is its ability to activate the memory-associated cAMP–PKA–CREB signaling pathway.²³⁵ Frequent exercise is also linked with decreased risk of viral and bacterial infections.²³⁶ Therefore, an increased susceptibility to infection as a result of insufficient exercise, which is shown to contribute to neurodegenerative pathology, may account for another mechanism by which exercise is linked with AD pathogenesis.²³⁷

Oral hygiene also appears to play a role in AD risk.²³⁸ More specifically, people with the gum disease known as periodontitis have persistent inflammation that is linked with AD.^238,239 For example, it has been demonstrated that serum antibody levels for periodontitis-associated bacteria are increased in AD patients, suggesting a potential link between AD and this peripheral chronic inflammation.²⁴⁰ In addition, levels of the pro-inflammatory cytokine TNF-α are increased in the serum of patients with periodontal disease.²⁴¹ Accordingly, TNF-α levels are also significantly increased in AD patients, suggesting a potential link between this specific inflammatory cytokine resulting from oral infection and the propagation of AD brain pathology.^241–243

Exposure to pathogens in the environment has also been shown to influence AD risk. Mounting evidence has produced a consequential list of bacteria, viruses, and fungi found in the AD brain, with some even directly associating with Aβ plaques and influencing tau phosphorylation.^90,244–247 Moreover, it has been reported that AD patients experiencing acute systemic inflammatory responses characterized by increased levels of TNF-α exhibit a two-fold increase in cognitive deficits.²⁴⁸ In fact, patients with sepsis present with long-term cognitive deficits, hippocampal atrophy, and are at a higher risk for developing dementia.^249,250

TBI has emerged as a major non-genetic risk factor for developing AD and other neurodegenerative disorders later in life. TBI not only significantly increases one’s risk of developing AD, but has also been linked to earlier onset and more aggressive disease progression.^251–254 More specifically, findings from a number of studies demonstrate that moderate to severe brain injury can increase the risk of developing AD by between 2.3- to 4.5-fold.^255–257 These numbers increase exponentially with repetitive brain injuries and interactions with genetic risk factors.²⁵⁸ For instance, possessing one allele of APOE ε4 doubles the risk of developing AD, whereas this risk increases by 10 fold for individuals who carry one copy of the APOE ε4 allele and have a history of brain injury.²⁵⁹ One mechanism thought to underlie the link between TBI and AD is BBB breakdown. TBI-induced vascular shear stress has been shown to appreciably disrupt BBB integrity.²⁶⁰ Cerebrovascular injury induced by TBI and AD has been postulated to be another pathological loop in which BBB breakdown propagates AD pathology and the presence of AD pathology can also incite the loss of BBB integrity.²⁶⁰ For instance, 30% of brains from TBI patients have Aβ plaques, and TBI can induce tau phosphorylation and Aβ aggregation.^251,261,262 The accumulation of Aβ and tau can then promote BBB breakdown to initiate the cycle once again.²⁶⁰ These findings suggest a link between this inflammatory brain injury and AD progression.

An additional mechanism by which brain injury and neurodegenerative disease interact involves the (re)discovery of the meningeal lymphatics in the CNS.²⁶³ In particular, the role of the meningeal lymphatics in neuroinflammation and Aβ accumulation may provide another compelling angle to the relationship between TBI and AD.^264,265 For instance, TBI has recently been shown to compromise meningeal lymphatic function, namely its ability to drain inflammatory byproducts to lymph nodes following injury.²⁶⁶ Interestingly, meningeal lymphatic drainage defects are known to significantly increase meningeal Aβ load and to provoke memory deficits in 5xFAD AD mice.²⁶⁵ Therefore, meningeal lymphatic dysfunction may provide another explanation for the association between AD pathogenesis and brain injury, with promising clinical implications that require further investigation.

BBB permeability is also known to be influenced by gut microbiota composition.²⁶⁷ Indeed, changes in intestinal microbiota landscape have been shown to have prominent effects on the brain either directly via production of microbial products that shape BBB function or indirectly through the modulation of peripheral immune responses.^267,268 Increased gut permeability allows for bacteria, bacterial toxins, and LPS to enter the bloodstream and has been coined “leaky gut.”^269,270 This leakiness is instigated by chronic inflammation and gut dysbiosis, which can contribute to systemic inflammation and the decreased BBB integrity associated with aging and AD.^229,269,271 BBB breakdown can ultimately lead to impaired removal of toxic species from the CNS and entry of pathogenic mediators from the periphery.^269,271 Interestingly, compared to WT controls, APPPS1 mice have altered bacterial species in the gut, including higher levels of the bacteria B. thetaiotaomicron.²²⁹ In these studies, it was also shown that modulating the gut microbiota landscape with probiotics and exercise treatment can improve cognitive functions in APPPS1 mice.²²⁹

In addition to rescuing memory, the modulation of microbiota composition upon fecal transfer from WT mice to ADLP^APT AD mice has been shown to attenuate Aβ load, tau pathology, and microgliosis.²⁷² Antibiotic cocktail treatment to alter microbiota species in the gut of APPPS1–21 AD mice has likewise been reported to reduce Aβ load and microglial activation in males specifically.²⁷³ Accordingly, cognitively impaired patients with high levels of Aβ also have altered gut microbiota populations. Interestingly, this skew in microbiota composition seen in cognitively impaired individuals is commonly characterized by the outgrowth of commensal microbes that have been linked with enhanced pro-inflammatory cytokine production.²⁷⁴ Dietary changes in patients with MCI also appears to alter the microbiota population and their association with Aβ and p-Tau levels in the CSF.²⁷⁵ These data convey how peripheral inflammation induced by dysbiosis in the gut can have broad clinical applications for mitigating AD risk and pathology in the future.

Sleep deficits have long been associated with cognitive decline.²⁷⁶ In fact, many AD patients present with shifts in their circadian rhythm.²⁷⁷ However, research also shows that problems with sleep can precede neurodegeneration and increase the risk of developing AD.^277,278 Notably, Aβ and tau levels are known to be elevated in the CSF and brains of sleep-deprived individuals.^279–281 In healthy patients, levels of Aβ shift, with an increase during the day and a marked decline during sleep.²⁸² However, following Aβ aggregation, this alignment with the sleep-wake cycle dissipates.²⁸² The half-life of Aβ more than doubles with age, likely allowing for improper protein folding and aggregate formation acting as a magnet for extracellular Aβ and preventing its transport into the CSF.²⁸³ Impaired microglial clearance of Aβ likely contributes to this enhancement in Aβ half-life. Deficits in sleep itself shift microglia to an activated state and aged phenotype, characterized by impaired Aβ clearance.²⁸⁴ Additionally, sleep disturbances increase the risk for inflammatory disease, which also has the potential to prime microglia to contribute to their impaired function with aging.^224,285

The disparity between female and male prevalence of AD has also sparked research to distinguish the pathways underlying this sexual dimorphism. The increase in incidence of AD in females is thought to involve females’ dramatic decrease in estrogen levels during menopause, which is known to impact mitochondrial metabolism.^286–288 In line with this hypothesis, estrogen replacement therapy (ERT) has been shown to significantly reduce the risk of developing AD.²⁸⁹ The decrease in testosterone in males occurs as a gradual decline, and therefore, it is believed to have a less detrimental effect on mitochondrial function.²⁸⁷ The decrease in sex hormone levels, characteristic during aging, affect mitochondrial metabolism by decreasing mitochondrial energy production and calcium efflux while increasing oxidative stress and reactive oxygen species (ROS) release.^287,290 The surge of oxidative stress and ROS amplifies inflammatory responses.²⁹¹ Accordingly, the release of ROS can trigger NLRP3 inflammasome activation and lead to downstream proinflammatory cytokine production, pyroptosis, and ASC speck seeding of Aβ spread.^{201,202,213,216,291,292} Some AD researchers are in favor of a “mitochondrial cascade hypothesis” in which mitochondrial dysfunction precedes AD pathogenesis by stimulating an increase in Aβ production and leading to the phosphorylation of tau.²⁹³ In a feed-forward mechanism, the AD pathologies themselves also propagate mitochondrial dysfunction and inflammatory ROS production which is consistently found at higher levels in AD patients.^294,295 This has led to great interest in AD therapeutics that promote mitophagy to eliminate impaired mitochondria.²⁹⁶ In fact, boosting mitophagy has been shown to attenuate cognitive deficits and promote microglial phagocytosis of Aβ while decreasing neuroinflammation in APPPS1 AD mice.²⁹⁷

An additional sex-specific factor involved in AD is the sexual dimorphism of microglia. Female microglia display distinct morphology, decreased density, and altered function.²⁷ It has also been demonstrated that male and female microglia in mice exhibit distinct transcriptional profiles that are retained when transplanted into the opposite sex.²⁹⁸ This suggests that the sex-specific microglia transcriptome does not rely on maintenance by sex-steroids. However, this does not exclude the possibility that the sex-steroids have a permanent effect on microglia during development. In fact, the development of microglia, as measured by gene expression profiles, occurs at disparate paces between the sexes.²⁹⁹ In turn, there is also a sex-specific divergence in the functionality of microglia in mice. For example, using a 5xFAD AD mouse model, it has been shown that female APOE ε4/5xFAD mice exhibit decreased microglia-mediated compaction of Aβ-plaques, decreased Trem2 expression, and increased plaque burden compared with male APOE ε4/5xFAD mice.³⁰⁰ Such studies suggest that sex-based differences in microglial response might help to explain the higher rates of AD in females.