Apolipoprotein E in lipid metabolism and neurodegenerative disease

Abstract

The dysregulation of lipid metabolism has emerged as a central component of many neurodegenerative diseases. Variants of the lipid transport protein, Apolipoprotein E (APOE), modulate risk and resilience for several neurodegenerative diseases including late-onset Alzheimer’s disease. APOE variants directly alter the lipid metabolism of cells and tissues and have been broadly associated with several other cellular and systemic phenotypes. Targeting APOE-associated metabolic pathways may offer opportunities to alter disease-related phenotypes and consequently, attenuate risk for and impart resilience against multiple neurodegenerative diseases. Here we review the molecular, cellular, and tissue-level alterations to lipid metabolism that arise from different APOE isoforms. These changes to lipid metabolism could help us understand disease mechanisms and tune neurodegenerative disease risk and resilience.

Lipid metabolism is a central player in neurodegenerative diseases

Lipids serve vital functions as membrane components, scaffolding molecules, and signaling molecules in every cell. Since the brain is the second-most lipid-rich organ [1,2], it is unsurprising that altered lipid homeostasis plays an important role in neurodegenerative diseases. In fact, the first described case of Alzheimer’s disease (AD) included the observation of aberrant lipid accumulation within glia [3]. Lipid dysregulation is also implicated in other late-onset neurodegenerative diseases like Parkinson’s disease (PD) and Lewy Body Dementia (LBD) [4-6]. Mutations in lipid-processing pathways also cause severe early-onset neurodegenerative diseases such as Tay-Sachs disease, Batten’s disease, Sandhoff disease, and Neimann-Pick disease [7]. Changes in lipid homeostasis have been linked to neuroinflammation, a central element of the pathology of many neurodegenerative diseases [8,9]. Several genetic risk factors for common neurodegenerative diseases like AD, PD, and LBD include lipid metabolism genes like APOE, CLU, INPP5D, ABCA7, TREM2, PLCG2, and GBA [10-12]. In this review, we detail not only how APOE variants perturb cellular and systemic lipid metabolism, but also how these perturbations may underlie and modulate risk for many neurological diseases.

APOE is a ubiquitous apolipoprotein

The human gene APOE encodes Apolipoprotein E (APOE), a ubiquitous lipid transport protein that binds a variety of lipid species, including cholesterol, phospholipids, and triglycerides in lipoprotein particles [13]. APOE transports lipids into cells via different cell surface receptors. Its mechanism of action with the low-density lipoprotein receptor (LDLR) is well described [14] while novel APOE receptors are emerging [15,16]. APOE is present in both the periphery and the central nervous system (CNS). In the brain, it is the most abundant lipoprotein as quantified by mass spectrometry [17].

In the periphery, APOE is primarily produced by hepatocytes and circulates in plasma [18]. In the CNS, APOE is mostly expressed in glia, which are key cellular mediators of metabolic homeostasis. Astrocytes, which modulate neuronal communication and metabolic health, produce the most APOE in the CNS, followed by microglia, the primary resident immune cells in the brain [19]. Other cell types in both the periphery (like macrophages and adipocytes) and CNS (like oligodendrocytes and pericytes) have been shown to produce APOE as well [20,21]. Although neurons do not produce an appreciable amount of APOE under basal conditions, studies have noted that neuronal stress and injury induce APOE expression [22-24].

APOE alleles modify disease risk

In humans, APOE exists in three major alleles – APOE2, APOE3, and APOE4 – which differ from each other at two amino acid positions (Figure 1, Box 1). APOE3 (rs7412 C / rs429358 T; C112 / R158) is the most common allele (allelic frequency of 78% in Caucasian populations) and is a neutral allele with respect to neurodegenerative disease risk [25]. APOE4 (rs7412 C /rs429358 C; R112 / R158) has an allelic frequency of 14% in Caucasian populations and is the strongest genetic risk factor for late-onset Alzheimer’s disease (LOAD). LOAD risk increases with APOE4 gene dose [26,27]. In contrast, APOE2 (rs7412 T / rs429358 T; C112 / C158) is present at 8% in Caucasian populations. APOE2 carriers display decreased risk for LOAD. The APOE alleles have strong associations with the key pathological hallmarks of LOAD, amyloid-beta and phosphorylated tau [28]. Although the association between APOE alleles and LOAD risk and protection exists across different ancestral backgrounds, the strength of the association varies (Box 2) [29,30]. Deep sequencing efforts and small familial studies have uncovered rare coding variants of APOE [31,32]. Some of these, like the Christchurch (rs121918393 ; R135S), Jacksonville (rs199768005; V236E), and R251G (rs26760666) variants [33-37], also have strong associations with resilience to LOAD while many others are of unknown clinical significance with respect to neurological diseases.

Figure 1. Numerous nonsynonymous APOE mutations modify neurological disease risk.

The mature APOE protein consists of two primarily alpha-helical domains (light grey sections), connected by an intrinsically disordered hinge region (dark grey bar; residues 200-215). The N-terminal domain contains the low-density lipoprotein receptor binding site (diagonal stripes; residues 136-150) while the C-terminal domain contains sites for lipid binding (dark grey; residues 244-272). In relation to APOE3 (C112; R158), risk mutations are depicted in red, protective mutations in blue, and mutations of unknown neurological effect are in grey. Dashed lines indicate predicted or mild directional effects. Diamond-headed pins reflect variants on an APOE4 (C112R) background. This diagram was adapted from data in the Alzforum APOE mutations database. Abbreviations: fs represents a frameshift mutation, ChC stands for Christchurch, delins is a deletion-insertion mutation.

BOX 1: Evolution of the APOE gene.

Although many vertebrate species have an APOE gene, current sequencing data suggests that, while it is polymorphic in humans, it is monomorphic in closely-related nonhuman primates [161]. In fact, APOE4 is the ancestral human allele with closely related variants present in non-human primates. Given that many neurodegenerative diseases occur late in life (post-reproductively), human APOE4 is unlikely to be under negative reproductive selection pressure. APOE3 and subsequently APOE2 evolved less than 200,000 years ago [162]. It is postulated that APOE2 and APOE3 were selected for since they confer advantages to changing lifestyles and diets – especially those including higher dietary lipids due to meat consumption [163]. The varied ability of different APOE isoforms to bind and internalize lipids could provide a molecular correlate for this hypothesis.

Despite its effects on disease risk, APOE4 is still maintained as the second most common allele in humans. Multiple studies have suggested that APOE4 may confer selective immune advantages early in life [164-167]. Given the intimate relationship between lipid metabolism and immune function, perhaps this connection occurs through APOE alleles impact on lipids [77,168,169]. Understanding the evolution of allelic diversity will better illuminate APOE’s effect on lipid metabolism and disease risk, at different life stages and ancestral backgrounds.

BOX 2: Effects of ancestry and sex on disease-associations of APOE alleles.

Most research on APOE alleles and LOAD genetics has been performed on Northern-European populations. Smaller studies in populations with diverse ancestral backgrounds have revealed that APOE4 allele frequency differs among different populations. Thus, while APOE4 is found in 14% of Caucasian American individuals, it is found in up to 40% of African Americans, 37% in Oceania, and 26% in Australia. In southern Asia and Europe the APOE4 allelic frequency is less than 10% whereas in Northern Europe it rises up to 25% [29,170-173]. The epidemiological effects of APOE alleles also vary by population. For example, in Korea, Japan, and Japanese-American populations, APOE4 confers a greater risk for LOAD than in Caucasian population [30]. In contrast, for Native Americans, Hispanic Americans, African Americans, and those with African ancestry, APOE4 is associated with lower LOAD risk than in Caucasian-American populations [30,174-177]. A recent study in a Chinese population demonstrated that APOE3 is more protective than APOE2 [178]. Some of these ancestry-specific effects have been attributed to the APOE haplotype [174,176]. One recent study has discovered a novel locus (19q13.31) that may contribute to the attenuation of APOE4-mediated AD risk in African Americans [179]. The diversity in epidemiological data demonstrates that APOE4-mediated risk can be mitigated, and discovery of the underlying mechanisms can present ancestry-specific strategies for therapeutic intervention.

Sex differences can also impact the association of APOE alleles and LOAD risk. Females are nearly twice as likely to be diagnosed with AD than males [180]. Although overall there are no clear clinical differences in AD incidence between male and female APOE4 carriers, a recent study demonstrated that the susceptibility of female APOE4 carriers may occur earlier than in males [181]. Further studies of APOE4 patients suggest that these sex differences may be related to different metabolite profiles [114].

Given the global rise of neurodegenerative disease and the variation in population demographics of APOE alleles, it is vital that future research addresses not only the genetic implications of ancestral background and sex but also the intersection of these variables on disease.

Although increased LOAD risk is the most notable consequence of APOE4, this allele also confers risk to many other neurological conditions, including LBD [11], poor recovery from traumatic brain injury [38,39], cerebral amyloid angiopathy [40-43], chronic traumatic encephalopathy [44], and chemotherapy-induced cognitive impairment [45-47]. APOE4, when compared to APOE3, is also associated with increased risk for non-neurological conditions like hypercholesterolemia, hypertriglyceridemia, cardiovascular disease, as well as severe and long COVID-19 [48-53]. Independent of its association with LOAD protection, APOE2, is also associated with longevity [54]. However, APOE2, when compared to APOE3, has been found to increase risk for hypercholesteremia, cardiovascular diseases, macular degeneration, and certain cancers [55-58]. In fact, a recent study of samples from the UK Biobank discovered that APOE alleles were associated with risk for a multitude of conditions including obesity, liver disease, chronic airway obstruction, and type-2 diabetes [29,59,60].

The extensive association between APOE alleles and disparate diseases suggests that APOE regulates general cellular processes. Recent studies have connected APOE alleles to differential effects on many central biological pathways like lipid homeostasis, glucose utilization [61], inflammation [62-64], cellular trafficking [65-67], and mitochondrial function [68]. These disruptions occur in many cell types and reveal the profound and pleiotropic effect of APOE variants on diverse aspects of cell biology. Given that lipid disruptions are present in multiple neurodegenerative diseases, that APOE alleles have distinct phenotypic consequences on multiple cell types, and that APOE binds and distributes lipids within and between cells, we hypothesize that APOE impacts cellular phenotypes and disease risk via its effects on lipid metabolism.

Perturbed lipid metabolism is one potential mechanism for how APOE influences neurodegenerative disease risk. Other hypotheses, including those that implicate APOE isoforms in altering protein aggregation and deposition, have had more years of exploration and evidence to support them. The hypothesis that effects on lipid metabolism may unify the pleotropic phenotypes of various APOE isoforms is emergent and requires more research to establish its primary mechanisms and causative effects on disease risk. Nevertheless, it is an attractive hypothesis given the broad influence of lipid metabolism on multiple areas of cell health; modulating lipid metabolism may be used to attenuate APOE-mediated risk for multiple diseases.

Researchers have studied the effects of APOE isoforms on amyloid-beta in the context of AD for many years [69-72]. Recent genetic and functional studies have allowed the field to recognize the impact of APOE isoforms on disease risk beyond amyloid-beta and AD. Mass spectrometry advances have enabled the quantitative profiling of many lipid species in various disease models. The confluence of the identification of genetic risk factors, the underexplored potential of targeting lipid state as a therapeutic option for neurodegenerative diseases, and new tools to study metabolites have all resulted in rising interest in understanding the biology of lipids in neurodegenerative disease. APOE, with its strong genetic and functional associations with many disease states, is at the heart of this interest. Although the research in this area is in its early stages, we discuss below what is known about how APOE isoforms impact lipid metabolism and neurodegenerative disease-associated disruptions in cellular systems, model organisms, and patient tissue.

Cellular effects of APOE isoforms on lipid metabolism

Impacts on Intracellular Metabolism

Multiple studies connect APOE4 and the accumulation of intracellular cholesterol (Table 1). Cholesterol accumulates in human induced pluripotent stem cell (iPSC)-derived APOE4 homozygous astrocytes and microglia [63,64,66], as well as in the endoplasmic reticulum of oligodendrocytes [20]. Both increased de novo synthesis [73] as well as reduced efflux [63,66] have been implicated in APOE4-associated cholesterol accumulation. A recent study on the epidemiologically-protective APOE3-Jacksonville variant suggested that its protection arises, in part, from its ability to promote cholesterol efflux [36]. Many studies have correlated changes in cholesterol biosynthesis, efflux, and retention to LOAD-associated phenotypes. Cholesterol accumulation in APOE4 astrocytes has been associated with decreased lysosomal function [66], increased pro-inflammatory cytokine production [66], increased matrisome signaling [73], impaired uptake of amyloid beta [64], and dysregulated cytoskeletal dynamics [66]. These correlations suggest that aberrant cholesterol levels may be central to neurodegenerative disease risk. Further work will be necessary to define a causal and mechanistic link between APOE4, glial cholesterol accumulation, and LOAD-associated pathologies.

Table 1.

APOE allele-associated metabolic phenotypes in different cell types and their relationships to neurodegeneration

Cell Type	APOE isoforms examined	Model System	Metabolic Phenotype	Relationship to neurodegeneration	References
Astrocytes	ApoE4, ApoE3	human iPSC-derived astrocytes	ApoE4 leads to decreased cholesterol efflux and increased intracellular cholesterol accumulation compared to ApoE3	Correlates with impaired lysosomal function, increased pro-inflammatory cytokine production, impaired amyloid-beta clearance, actin cytoskeletal defects	[64,66,73]
			ApoE4 leads to increased triglyceride accumulation and lipid droplet formation compared to ApoE3	Increased sensitivity to fatty acid lipotoxicity	[76]
			ApoE4 leads to decreased oxygen consumption and increased glycolysis compared to ApoE3	Correlates with pro-inflammatory astrocyte state	[196]
		primary astrocyte culture from humanized APOE mouse	ApoE4 leads to more lipid droplet accumulation upon lipid challenge compared to ApoE3	Decreased ability to oxidize excess fatty acids, increased sensitivity to lipotoxicity	[75]
	ApoE3, ApoE3-V236E	AAV-ApoE treated amyloid mouse model, primary mouse astrocytes	ApoE3-V236E results in greater cholesterol efflux and better lipidation than ApoE3	Correlates with reduced amyloid plaque load	[36]
Microglia	ApoE4, ApoE3	human iPSC-derived microglia	ApoE4 leads to triglyceride accumulation, lipid droplet formation, and cholesterol accumulation compared to ApoE3	Results in microglial activation, neuroinflammation, decreased neuronal activity	[63,73,76]
		primary microglia from humanized APOE mouse	ApoE4 leads to decreased oxidative phosphorylation, increased glycolysis compared to ApoE3	Associated with increased microglial reactivity and similar transcriptome to disease-associated microglia	[62]
Oligodendrocytes	ApoE4, ApoE3	human iPSC-derived oligodendrocytes	ApoE4 leads to cholesterol, triglyceride, diglyceride accumulation compared to ApoE3	Leads to impaired axon myelination	[20]
		human post-mortem tissue	ApoE4 leads to cholesterol accumulation, lipid droplet accumulation compared to ApoE3	Associated with impaired myelination	[20]

In addition to cholesterol, APOE4 homozygosity results in accumulation of triglycerides within astrocytes and microglia (Table 1). Many of these studies report the storage of triglycerides within lipid droplets. Lipid droplets are organelles that bud from the ER membrane into the cytosol, and are composed of a protein-functionalized phospholipid monolayer that encapsulates a neutral lipid core [74]. In astrocytes from humanized APOE4 mice as well as in iPSC-derived astrocytes and microglia, APOE4 induces the formation of triglyceride-rich lipid droplets [63,75,76]. Although it remains unclear why APOE4 glia accumulate lipid droplets, one study finds that APOE4 astrocytes exhibit a decreased rate of fatty acid beta-oxidation [75], suggesting a role for decreased lipid catabolism. Lipid droplets are commonly thought to buffer cells against cytotoxic lipids (reviewed in [74]). The presence of abundant lipid droplets in APOE4 glia under basal conditions may suggest that this buffering mechanism is impaired or overwhelmed, thus rendering APOE4 glia more sensitive to lipotoxic insults [76]. In both microglia and astrocytes, APOE4-induced triglyceride accumulation is associated with a proinflammatory state [62,63,73]. Lipid-associated proinflammatory states in microglia have also been implicated in neurodegenerative disease and aging models [77]. Current research links APOE4-associated triglyceride accumulation to disrupted cellular functions of stress tolerance and neuroinflammation. Further work will help understand how these disrupted cellular functions contribute to increased neurodegenerative disease risk.

APOE4 exacerbates AD-associated phenotypes (amyloid-beta and tau accumulation, vascular defects, neuroinflammation, behavioral phenotypes) in other non-glial human iPSC-derived cell types [20,73,78], and mouse models [79-82]. These include pericytes (part of the blood brain barrier; [21,83,84]), neurons [78,82], as well as neurovascular immune cells [80]. Although these diverse cell types are involved in neurodegenerative disease risk and progression, future work is needed to establish how changes to intracellular lipid metabolism impact their cellular function.

Effects on neuron-glia interactions

APOE isoforms can impact neuronal health through modulating the exchange of lipids between glia and neurons (Figure 2). Here we will examine impacts of APOE isoform on interactions between neuron and various glial cells—astrocytes, microglia, and oligodendrocytes.

Figure 2. APOE isoforms impact communication between glia and neurons.

(A) Astrocytes provide metabolic support to neurons by supplying neurons with lipids such as cholesterol and metabolizing lipid species secreted by neurons. These activities are impaired by APOE4. (B) APOE4, but not APOE3, impairs calcium signaling in both microglia and neurons. APOE4 microglia also accumulate triglycerides, exhibit decreased purinergic signaling, and secrete proinflammatory cytokines. (C) Oligodendrocytes myelinate neurons. APOE4 oligodendrocytes display lower viability and intracellular cholesterol accumulation which leads to decreased myelination capacity. This figure was created using Biorender.com.

Lipid metabolism mediates neuron-astrocyte interactions

APOE genotype has a strong effect on neuron-astrocyte interactions. Work in Drosophila and mouse primary culture models established the role of APOE in transporting toxic peroxidated lipids and excess triglyceride-rich lipid droplets from neurons to astrocytes for degradation [85-87]. Human APOE4 impairs the transport of toxic lipids as well as astrocytic lipolysis [85,87,88]. Deficient clearance of lipids is correlated with increased oxidative stress, mitochondrial dysfunction, bioenergetic defects, and hampered synaptic transmission in neurons [88]. Neurotoxicity can also be induced by reactive astrocytes, an important component of the neuroinflammatory response in many neurodegenerative diseases [89,90]. APOE and clusterin-coated saturated fatty acid-rich particles are the mediators of reactive astrocyte-induced neurotoxicity [91]. How different APOE variants impact this function is undetermined.

Neurons also require a large amount of cholesterol and depend on astrocytes and APOE for adequate supply [92,93]. As the primary lipoprotein in the CNS, APOE delivers cholesterol to neurons. Independently, alteration of neuronal cholesterol levels has been implicated in LOAD-associated tau and amyloid-beta pathology [94]. The role of APOE in distribution of cholesterol may offer a direct link between APOE4 and LOAD pathology.

Lipid metabolism mediates neuron-microglia interactions

The interactions between microglia and neurons are also modulated by APOE4-associated lipid dysregulation. TAG-rich lipid droplet accumulation was found to induce a proinflammatory state in microglia in mouse models [77]. APOE4 homozygous microglia that accumulate triglycerides and cholesterol display a proinflammatory phenotype, exhibit reduced homeostatic purinergic signaling, and impact neuronal activity [63]. Direct modulation of the microglial lipid droplet levels using acyl-coA synthase inhibitors impacts both microglial neuroinflammation and neuronal activity [63]. These findings mechanistically link perturbed microglial lipid metabolism and neuronal function. Further work in degenerative disease models will be needed to confirm whether these cellular-level modulation of lipids can impact disease progression.

Lipid metabolism mediates neuron-oligodendrocyte interactions

APOE4-influences myelination in cellular, mouse, and human tissue. Mixed and matched co-cultures of iPSC-derived neurons and oligodendrocytes of APOE3 and APOE4 homozygous genotypes identified that aberrant cholesterol accumulation in APOE4 oligodendrocytes resulted poor myelination of cocultured neurons [20]. This same study showed deficient myelination in APOE4 humanized mice compared to their APOE3 counterparts. Another study in humanized APOE mice suggested that APOE4 impacts the viability of the oligodendrocytes themselves [95]. Both outcomes were loss-of function effects and result in poor myelination of axons, which has been associated with poor neuronal transmission and connectivity. A third study in a APOE4 homozygous tauopathy mouse model identified a gain of toxic function effect of neuronal APOE4 in causing myelination defects as well as decreasing the number of oligodendrocyte and oligodendrocyte precursor cells [82]. Interestingly, these effects were specific to a tauopathy mouse model suggesting that the interaction between APOE4 genotype and tau pathology may have specific effects on myelination.

The studies described above have started to unravel the complex relationship between APOE isoforms, their effects on glial lipid metabolism and the consequent cell non-autonomous effects. Further mechanistic studies are needed to truly decipher how modulation of APOE-associated lipid defects impacts disease risk and progression.

APOE isoforms impact lipoproteins and lipid transport

In the CNS, APOE resides on high density lipoprotein (HDL) particles whereas in the periphery it binds both HDL and very low density lipoproteins (VLDLs) [96,97]. The nature of the HDL particles has been reported to impact neurodegenerative disease phenotypes. One study reported that high density lipoproteins (HDLs) increase vascular amyloid deposition in in vitro models and HDLs containing APOE exacerbate that effect [83]. Another study of human plasma lipoproteins reported lower APOE content within HDL particles in patients with at least one APOE4 allele [98]. Moreover, the same study observed that patients with lower levels of APOE in their plasma or within HDL particles performed worse on cognitive function tests. These and other studies suggest that lipoprotein particle composition may impact neurodegenerative disease phenotypes and further study is necessary to understand the mechanisms that underlie these correlations.

APOE is lipidated in the secretory pathway of the cell by a lipid efflux protein, ABCA1 [99,100]. The lipidation status of APOE isoforms is different, with APOE4 reported to be most poorly lipidated [101,102]. Consequently, APOE isoforms have been shown to alter lipoprotein size and type—APOE4 is associated with smaller HDL particles than APOE3, which, in turn, binds smaller particles than APOE2 [103-105].The lipid species bound in lipoproteins are dictated by the lipid content of the cell type of origin and the APOE isoform present [98]. APOE isoforms are also found to differentially regulate the levels of proteins involved in lipidation, revealing a cyclical relationship between lipidation and APOE genotype [19]. Deficient APOE4 lipidation may compromise the ability to export excess lipids from cells [106-108] and alter the amount and nature of lipids transported to neurons and the vasculature, thereby modulating essential neuronal functions like synaptic connectivity and vascular integrity [109-111]. Additionally, APOE4 protein levels are observed to be lower compared to APOE3 levels within and secreted from multiple cell types and in plasma [64,112]. This may further exacerbate impaired lipid efflux.

A few studies have characterized the plasma, CSF, and brain lipidome of APOE4 carriers compared to non-carriers. One such study of 58 individuals identified 6 plasma metabolites (including lysophospholipids and cardiolipin) that were downregulated in APOE4 carriers [113]. Another study of the plasma metabolome of APOE4 carriers identified a preference for aerobic glycolysis [61]. A larger study of over 1500 individuals also found a few metabolites, including many phosphatidylcholines, significantly associated with both APOE genotype and sex [114]. The variability between these studies is likely attributable to differences in the subjects profiled as well as in the metabolomic coverage achieved. Changes in peripheral metabolites could serve as risk or resilience biomarkers, however, more work must be done to understand how blood metabolites reflect CNS APOE-associated processes (Box 3).

BOX 3: APOE-dependent, peripheral effects on the CNS.

It has been long thought that the CNS and peripheral pools of APOE do not mix [182,183]. However, recent studies have revealed that the two pools influence each other. This may occur through direct leakage via a more permeable blood brain barrier [84] or through indirect effects to systemic metabolism and signaling. In humanized APOE knock-in mouse models and those crossed to tau models, knocking out peripheral APOE impacts brain pathology while restoring peripheral APOE in a knockout organism can rescue cognitive defects associated with APOE loss of function [184,185]. In another amyloid mouse model, removal of peripheral APOE had less of an effect on brain pathology [186]. Together, these mixed results suggest systemic crosstalk, perhaps through metabolite exchange, where peripheral APOE impacts brain pathology. However, how interactions between peripheral APOE and brain APOE affect pathology may be specific to which model is used.

Dietary changes have also been shown to impact the CNS in an APOE isoform-dependent manner. For example, in APOE3 but not APOE4 humanized mouse models, high fat diets have been shown to increase both gliosis and neuronal immediate-early gene expression [187]. In younger adults, obesity predisposes APOE4 individuals to greater cognitive decline when they get older [188]. On the contrary, in older adults, obesity in APOE4 carriers result in slower cognitive decline [189]. These and other studies reveal the complex interplay between diet and CNS impacts of APOE4 alleles.

Additional studies in mice have demonstrated that APOE genotype can influence the gut microbiome [190,191] and vice versa – that alterations in the gut microbiome can also influence the role of APOE4 in tau pathology [192,193]. The mechanisms of this association are sex-dependent and may operate through systemic metabolic alterations [194]. Similar observations have been seen in humans [195]. However, the specific effects on gut microbiota and their consequences on neurodegenerative disease pathology is an area ripe for future exploration.

APOE mediated lipid metabolism perturbations in the brain

While cell type-specific and pairwise interactions can shed light on mechanistic relationships between glia and neurons, tissue-level studies enable the investigation of the regional and systemic effects of APOE.

Studies profiling lipids in an APOE knockout mouse model revealed APOE as a key regulator of multiple lipid metabolic pathways, including accumulation of cholesterol esters in glia [115,116]. Transcriptomic and lipidomic profiling in humanized APOE targeted replacement mice revealed both cell type-specific, brain regional, and plasma differences in lipid metabolism [117,118]. These studies identified APOE4-associated decrease in free fatty acid levels, increases in many TCA cycle metabolites as well as plasma changes in both phosphatidylcholines and unsaturated fatty acids. Another study applied lipidomics to an amyloid mouse model harboring various humanized APOE alleles and identified key brain region-specific lipid changes that differentiate areas vulnerable to early AD pathology (like the entorhinal cortex) and those that are more resistant to amyloid pathology (like the primary visual cortex) [119]. In the entorhinal cortex, the authors discovered an APOE4-associated decrease in diacylglycerol levels concomitant with increases in sphingolipids and bis(monoacylglycerol)phosphates. Studies like these nominate lipids as potential markers of vulnerability or resilience of different cell types and brain regions. Much work remains to understand the neurological consequences of lipid profile changes and the mechanisms by which they contribute to disease risk, resilience, and selective vulnerability of brain regions.

CSF profiling in APOE-genotyped individuals revealed some genotype-specific metabolites and the influence of peripheral dietary lipids [120]. One metabolomic characterization of post-mortem brain tissue from both healthy individuals and AD patients reported that APOE4 resulted in detectable sterol and sphingolipid disruption specifically in AD patients [121]. Other metabolomic characterizations of post-mortem human brain tissue have failed to identify metabolites significantly associated with the APOE4 risk genotype [20,122], though they do hint at trends in increased cholesterol esters, unsaturated lipids, and a few sphingomyelin species. This finding suggests that the variability of post-mortem brain tissue samples requires much higher sample sizes, profiling of comparable brain regions, and control of acquisition conditions to generate reproducible genotype-specific lipid signatures.

Advances in mass spectrometry imaging (MALDI-MSI) techniques have enabled the spatial tracing of metabolites. Recent studies have employed a combination of omics techniques and MALDI-MSI imaging to reveal that APOE4 microglia in humanized mouse models show a significant metabolic shift contributing to disease pathology [62]. Studies like these pave the way for relating metabolic changes on a cell type and regional level to disease progression and pathology, and will help bridge the gap to better understand whole organism effects of APOE alleles (Box 3)

Targeting APOE-associated disruptions to lipid metabolism

Given the pronounced effects of APOE alleles on disease risk and lipid metabolism, modulation of lipid metabolism is an attractive prospect to attenuate risk and impart resilience to disease. Many of these genotype-targeted approaches are still in early, preclinical stages and require significantly more basic research to truly understand their underlying mechanisms and broader effects.

Depletion of APOE as a therapeutic approach has emerged from work in mouse models. In transgenic mice expressing human P301S tau and human APOE4, astrocyte-specific genetic depletion of APOE4 reduced phosphorylated tau deposition and tau-mediated neurodegeneration, reduced glial activation signatures, and rescued cognitive defects [123]. Another study using a similar model found that selective depletion of APOE4 from neurons likewise decreased phosphorylated tau deposition and neurodegeneration, improved oligodendrocyte progenitor cell-mediated myelin degeneration, and was associated with a shift of astrocytic and microglial populations from disease-associated to disease-protective profiles [82]. Taken together, these studies suggest that there are toxic gain-of-function mechanisms of APOE4 in multiple cell types that may have cell autonomous and non-autonomous effects. However, there are many other studies that point to the loss-of-function effects of APOE4, especially in lipid binding (reviewed above). It will be important to understand the relative balance between loss-of-function and gain-of-toxic-function in the context of neurodegeneration in human systems before APOE4 depletion can be a viable preventative strategy.

A clinical trial has recently begun to test whether adeno-associated virus (AAV)-mediated expression of APOE2 in the CNS can alleviate AD-associated metrics in APOE4 homozygous individuals with mild and moderate dementia (https://clinicaltrials.gov/ct2/show/NCT03634007). The initial phase of the trial is focused on evaluating adverse events and toxicity. The trial is based on preclinical evidence in mouse models that showed reduction in brain amyloid levels upon APOE2 AAV injection, especially at lower levels of preexisting pathology [124,125]. If successful, this trial would provide precedent that risk-to-resilience conversion can be accomplished through exogenous genotype supplementation.

The best-explored avenue to modulate APOE-associated lipid metabolism is by activating retinoid X receptor (RXR)/liver X receptor (LXR) signaling. The RXR/LXR signaling axis controls the biosynthesis of lipids through the expression of many genes, including APOE and ABCA1 [126,127]. Activation of the RXR/LXR pathway can promote APOE4 lipidation, partly compensating for its inherently deficient lipidation [128]. RXR/LXR agonists like bexarotene (an FDA approved drug for certain cancers) have shown promise in reversing APOE4 lipidation defects and consequent cognitive defects in mice [129,130]. However, bexarotene has shown disparate results in different mouse models [131-133] suggesting that its mechanism of action is complex and may result in pleiotropic effects [134]. Two clinical trials explored the effect of bexarotene on both healthy individuals and AD patients [135,136]. These small trials showed no clear cognitive benefit, mixed results regarding changes to pathology, and increased the risk of adverse events. Other RXR/LXR agonists have also proceeded to phase 1 clinical trials but many have not progressed further due to adverse effects. These include increased triglyceride levels, decreases in neutrophils in humans, and hepatic steatosis in mouse models [137,138].

Peroxisome proliferator activated receptors (PPARs) are another class of nuclear receptors that form heterodimers with RXRs to drive expression of several lipid-associated genes, including ABCA1 and APOE. Activation of PPAR-gamma signaling with the agonist rosiglitazone has been associated with an improvement in learning and memory deficits and cerebrovascular function in several transgenic mouse models of AD [139-141]. However, multiple clinical trials to investigate safety and efficacy of rosiglitazone and pioglitazone therapy showed no clear benefit to either APOE4 carriers or APOE4 non-carriers while inducing the detrimental side effect of oedema [142-144].

A more nuanced approach to targeting APOE lipidation is via the activation of ABCA1, the transport protein responsible for lipidating APOE. For example, CS-6253 is an ABCA1 agonist that mimics the C-terminal domain of APOE itself. Such ABCA1 agonists have shown reduced AD pathology and cognitive deficits in an APOE4 mouse model [145]. However, ABCA1 agonists also alter the lipoprotein distribution in the plasma. Studies on CS-6253 in rodents and primates have shown effects in altering peripheral lipoprotein distribution and non-significant trends in CSF amyloid-beta 42 reduction, but do not impact CSF lipoproteins [146]. These approaches are still in preclinical stages. They do offer more precise alternatives to RXR/LXR and PPAR-gamma modulation, both which increase efflux through ABCA1, but result in unwanted hepatic lipogenesis.

Given the excess cholesterol accumulation in multiple APOE4-associated contexts, modulation of cholesterol has also been explored as an APOE4-targeted therapeutic. In a recent study, hydroxypropyl beta cyclodextrin, a cholesterol-depleting drug, was fed to APOE4 mice resulting in fewer myelination defects and more favorable performance on behavioral tests [20]. Many trials have been performed to evaluate whether statins (which inhibit cholesterol biosynthesis) ameliorate APOE4-associated phenotypes. APOE genotype-stratified trials revealed that APOE4 carriers show some reduced disease progression and cognitive decline when on statins compared to other APOE genotypes [147], further highlighting the importance of the APOE genotype-lipid metabolism axis in neurological disease.

Other potential therapeutic candidates have emerged from cellular studies. Choline supplementation was shown to reduce lipid burden in APOE4 astrocytes [76]. Choline is a readily available nutritional supplement with few side effects even when taken at high doses (https://ods.od.nih.gov/factsheets/Choline-Consumer/). It also displays general neuroprotective effects in multiple mouse model studies [148-151] and has shown benefit in humans as well [152]. The benign nature of choline supplementation presents an attractive option as an APOE-specific preventative intervention. Although choline has been tested as a general AD therapeutic with mixed results [153-155], many of these studies were not conducted on APOE genotype-stratified patients. Those trials that did stratify on APOE genotype did note greater efficacy in APOE4 carriers [156]. Further work on genotype-specific mechanisms of neuroprotection needs to occur to establish choline as a promising modulator of APOE4-mediated risk.

Given that APOE4 is a disease risk factor, not a causative gene, a genotype-specific preventative treatment has to be highly tolerable and have a biomarker [157,158] that can indicate prevention of disease onset or progression.

Concluding Remarks & Future Directions

Over 100 years ago Alois Alzheimer initially observed glial lipid accumulation in a post-mortem AD brain [3]. Today, we recognize the central role lipid metabolism plays in APOE-mediated risk and resilience for neurodegenerative diseases. Studying the role of APOE alleles in neurodegenerative diseases independent of disease-specific pathology can enable the discovery of common nodes and targetable pathways for many conditions. Cell biology studies in cultured glia have revealed cell type-specific mechanisms and consequences of APOE4-associated lipid disruptions. Co-culture studies have revealed the role of APOE isoforms in transferring lipids between astrocytes and neurons and shaping microglial states. However, much work remains to bridge the gap between these mechanistic studies and understanding the consequences of modulating lipid state in complex tissues like the brain. Although existing work in brain tissue has begun to characterize lipidomic changes in response to APOE genotype, much work is still needed to connect modulation of specific lipid metabolic pathways to pathology or cognitive outcomes. Finally, being able to relate findings in model systems to human tissue and physiology is crucial to substantiate disease relevance.

In addition to the continuation of studies like those reviewed above, we propose select promising lines of investigation and outstanding questions for the field (see Outstanding Questions).

Outstanding Questions.

How do APOE alleles with different sequences give rise to changes in lipid metabolism in different cell types and in organisms as whole?

Does ancestral background and sex impact the effects of APOE alleles on lipid metabolism?

What are the functional consequences of APOE allele-induced changes to lipid metabolism?

What are genetic and chemical modifiers of lipid metabolism that may present viable preventative strategies?

In past years, various research groups have identified novel protective alleles of APOE. Identifying the effects that protective alleles have on lipid metabolism may hold the key to modulating cellular lipid state to promote disease resilience. We look forward to seeing the expansion of understanding in this area.

APOE has been studied in the context of perturbations to peripheral lipid metabolism by the cardiovascular disease research field for years. Fostering communication between the cardiovascular disease and the neurodegenerative disease fields may provide insights into how APOE isoforms regulate disease risk in a fundamental way. Cross-disciplinary meetings and grant mechanisms could foster such communication and further fundamental research into the functional consequences of APOE-mediated lipid perturbations. This approach could have particular significance to the area of vascular dementia.

Technical advances in studying metabolites have enabled researchers to identify how the various APOE alleles modify risk for neurodegenerative disease through actions on cellular and organism-wide metabolism. Metabolomics approaches, including spatial metabolomics, are only improving in their sensitivity and specificity of detection [159]. Further, high-powered lipidomic characterization of large post-mortem tissue cohorts will help distinguish genotype-specific lipid signatures from the natural sample-to-sample variability. As these approaches are widely adopted, standardization of pipelines and preparation methods is important to allow data to be comparable across technical platforms. These platforms can accelerate the identification of new drug targets, engagement of targets by drug treatments, and mechanisms of actions of novel modulators [160].

Ultimately, changes in lipid metabolism are central to the modulation of disease risk by APOE alleles. Many studies are just uncovering the key determinants of these metabolic changes. Further work is needed to substantiate causal links between APOE-associated lipid changes and disease risk or resilience via chemical or genetic modulation of relevant lipid species. A comprehensive understanding of the cellular regulators of lipid metabolic disruptions will be central to targeting these nodes for therapeutic or preventative benefit.

Highlights:

Human APOE alleles modify lipid metabolism and impact risk for multiple neurodegenerative diseases.

APOE4 is associated with triglyceride and cholesterol accumulation in glia, modulating their immune reactivity.

APOE4 is poorly lipidated compared to APOE3 and prevents efficient transport of lipids between glia and neurons. This impacts neuronal homeostasis and survival.

APOE4 alters brain lipid metabolism in mouse models and human tissue.

Targeting APOE-associated changes to lipid metabolism could form the basis of genotype-specific preventative strategies for a broad range of neurodegenerative diseases.