Peripheral Versus Central Nervous System APOE in Alzheimer’s Disease: Interplay across the Blood-Brain Barrier

Abstract

The apolipoprotein E (APOE) ε4 allele has been demonstrated as the preeminent genetic risk factor for late onset Alzheimer’s disease (AD), which comprises greater than 90% of all AD cases. The discovery of the connection between different APOE genotypes and AD risk in the early 1990s spurred three decades of intense and comprehensive research into the function of APOE in the normal and diseased brain. The importance of APOE in the periphery has been well established, due to its pivotal role in maintaining cholesterol homeostasis and cardiovascular health. The influence of vascular factors on brain function and AD risk has been extensively studied in recent years. As a major apolipoprotein regulating multiple molecular pathways beyond its canonical lipid-related functions in the periphery and the central nervous system, APOE represents a critical link between the two compartments, and may influence AD risk from both perspectives. This review discusses recent advances in understanding the different functions of APOE in the periphery and in the brain, as well as highlighting several promising APOE-targeted therapeutic strategies for AD.

Introduction

Apolipoprotein E (APOE) is a 34 kDa glycoprotein that is synthesized primarily in the liver and in the brain. ApoE is incorporated into lipoprotein particles in the plasma in the periphery, as well as in the cerebrospinal fluid (CSF) and in the interstitial fluid (ISF) of brain parenchyma in the central nervous system (CNS) [1, 2]. The canonical function of apoE is in transporting cholesterol and other lipids, mediated by cell surface apoE receptors, although APOE also plays important roles in immunomodulation, synaptic plasticity, signal transduction, and proteostasis [3, 4].

The APOE gene is polymorphic in humans, consisting of three alleles (ε2, ε3, and ε4), which leads to the production of three distinct isoforms of APOE protein: APOE2, APOE3, and APOE4. The three APOE isoforms differ by one single amino acid residue at position 112 or 158: APOE2 has cysteine residues at both positions, APOE3 has a cysteine residue at 112 and an arginine residue at 158, and APOE4 has arginine residues at both positions [5, 6]. These differences alter the structure and physiological function of APOE. APOE2 may elicit type III hyperlipoproteinemia and APOE4 increases the risk for hypercholesterolemia and atherosclerosis [7].

Interest in APOE in the CNS escalated greatly when it was discovered that the APOE4 allele is associated with a significant increase in the risk for late-onset Alzheimer’s disease (AD), while the presence of the APOE2 allele protects against the development of AD [8–10]. The allele frequency of APOE4 is approximately 15% in the general population; however, it is enriched to ~ 40% in AD patients. APOE3, the most common allele, has a frequency of approximately 77%, and is considered to be neutral in regards to neurodegeneration, while the APOE2 allele is less common in the general population (~ 8% allele frequency) [8–10].

Circulating lipoproteins are unable to cross the blood brain barrier; however, small high-density lipoprotein (HDL) particles can enter the brain, stimulating interest in identifying the relevance of plasma HDL and associated apolipoproteins in brain function [11, 12]. This review aims to capture recent findings on the role of both peripheral and brain APOE in the pathogenesis of AD and highlight emerging APOE-targeted therapeutics for AD.

Two pools of APOE

The most compelling support for two distinct pools of APOE in the human body came first from a liver transplantation study in 1991 [13]. It was found that in patients after liver transplantation, the APOE isoform present in the plasma almost completely converted to that of the donor whereas the APOE isoform in the brain did not change. This study demonstrated that APOE in the periphery is primarily produced by the liver and that the protein does not cross the blood brain barrier (BBB) [13]. In the brain, APOE is constitutively produced by glial cells (astrocytes, oligodendrocytes, and microglia), pericytes, vascular smooth muscle cells, choroid plexus, and limitedly by neurons under stress conditions [14–16]. In the periphery, although the major source of APOE is from hepatocytes in the liver, APOE is also expressed by monocytes and macrophages [1, 17, 18]. Concentrations of APOE differ highly between the two compartments; APOE levels are around 50–70 μg/mL in the plasma, whereas they are in the range of 3–10 μg/mL in the CSF [19, 20].

The regulation of APOE expression is tissue- and cell-specific and is controlled by several transcription factors and regulatory elements, as recently reviewed [18]. The expression of APOE can be induced by environmental factors including cytokines, hormones, or lipids. In the periphery, APOE expression in the hepatocytes and macrophages requires the binding of specific distal enhancers, as well as common proximal enhancer elements, to the promoter region. Notably, although the amount of APOE expression is low in macrophages, local production of APOE by macrophage exerts dramatic protection against hypercholesterolemia-induced atherosclerosis [21, 22]. In the CNS, glial APOE expression is regulated by specific transcriptional factors through distinct binding sites on its promoter [18]. Neuronal APOE expression is uniquely regulated, in which an intron-containing APOE splice variant is produced but not translated; upon injury, this splice variant is further processed to produce normal mRNA for translation to protein [23]. In addition, recent studies have indicated the expression of APOE is also subjected to certain epigenetic controls [18].

The posttranslational modification of APOE in the periphery and CNS is also differentially regulated. Newly produced APOE follows the classical secretory pathway. Glycosylation and sialyation of APOE occurs in the Golgi and are uniformly present in APOE secreted from hepatocytes and macrophages [24]. However, APOE in plasma typically lacks carbohydrate molecules [18, 25]. In contrast, APOE in CSF remains heavily glycosylated and sialylated, consistent with the form of APOE secreted from astrocytes. The sialic acid groups reduce the isoelectric point of apoE, and glycosylated apoE is less soluble in Tris buffered saline (TBS) containing the detergent Triton-X100 than in TBS alone, suggesting that this post-translational modification may influence the ability of APOE to bind lipids [25]. APOE4 was more commonly sialylated, and more commonly found in the TBS-soluble fraction, and less commonly in the fraction containing detergent, than APOE3, suggesting that this post-translational modification may underlie the reduced lipid binding observed for APOE4.

Recent studies of APOE levels in CSF, brain parenchyma, and plasma of individuals with different APOE genotypes have provided further evidence of distinct peripheral and central APOE pools in the context of AD. It has been shown that the correlation between CSF and plasma APOE levels is very low [26]. Importantly, APOE4 carriers have a decreased level of APOE, whereas APOE2 carriers have an increased level of APOE in CSF and plasma compared with APOE3 carriers [26]. This study further showed that CSF but not plasma APOE levels are associated with CSF Aβ42 levels. However, another study found no differences in CSF APOE levels between APOE genotypes but showed that APOE4 carriers exhibit significantly decreased plasma APOE levels, resulting from a specific decrease in APOE4 concentrations [27], and suggested that the life-long decrease in plasma APOE may predispose APOE4 carriers to AD. Another recent study reported a robust imbalance in APOE isoform distribution in APOE3/APOE4 heterozygous individuals [28]. In CSF and cortical tissue, APOE4 levels were higher than APOE3 levels, whereas in plasma the relationship was reversed; CSF and plasma APOE4/APOE3 isoform ratios were negatively correlated between paired plasma and CSF samples. These observations indicate that APOE in the periphery and CNS exhibits different isoform dependent turnover rates. Plasma APOE turnover is isoform-dependent with a rank order of APOE4 > APOE3 > APOE2. In contrast, CNS APOE turnover is three- to six- fold slower and is isoform independent. The study also showed that the level of APOE in CSF increases with age and that amyloid deposition is associated with further elevation of APOE3, but a decrease of APOE4 in the CSF [28]. Further, APOE levels in the CSF were associated with CSF amyloid levels, whereas plasma APOE levels lacked this correlation. However, other recent large studies have shown that low plasma/serum APOE levels are associated with an increased risk of AD/dementia, independent of APOE genotype, in general population [29–31].

Taken together, these studies demonstrate that peripheral and CNS APOE are separate pools and are distinctly regulated. Emerging evidence indicates peripheral APOE, as well as CNS APOE, influences the progression of AD directly or indirectly through multiple pathways as shown in Fig. 1 and discussed below.

Fig. 1.

A schematic to illustrate that APOE in the periphery and in the brain regulates multiple molecular pathways pertinent to the pathogenesis of AD. See the main text for details. Aβ: amyloid-β; ABCA1: ATP-binding cassette transporter A1; APOE or E: apolipoprotein E; BBB, blood-brain barrier; HDL, high-density lipoprotein; VLDL: very low-density lipoprotein; LDLR, low-density lipoprotein receptor, LRP1, low-density lipoprotein receptor-related protein 1; TREM2: triggering receptor expressed on myeloid cells 2.

APOE in cholesterol/lipid metabolism

APOE binds primarily to the low-density lipoprotein (LDL) receptor family, which plays crucial roles in cholesterol homeostasis. APOE3 and APOE2 are more selective to HDL, while APOE4 preferentially binds to triglyceride-rich, very low-density lipoproteins (VLDL), leading to downregulation of LDL receptors and thus reduced clearance of LDL and increased plasma cholesterol levels [7]. These differences in affinity lead to increased risk of atherosclerosis and stroke among APOE4 carriers, as well as increased risk of type III hyperlipoproteinemia in those carrying APOE2, especially in homozygous APOE2 individuals because of defective binding of APOE2 to LDL receptors [7]. Some studies indicate that APOE2-induced type III hyperlipoproteinemia only occurs in the presence of other conditions such as diabetes, hypothyroidism, and obesity, which leads to fewer LDL receptors, limiting the ability of APOE2 to modulate cholesterol clearance [7]. The critical importance of APOE in cholesterol/lipid metabolism in the periphery was demonstrated by the phenotypes of ApoE-knockout mice, which include the spontaneous development of severe hypercholesterolemia and atherosclerosis [32, 33].

Cholesterol homeostasis is integral to normal brain function. While the brain comprises approximately 2% of the total body weight, as much as 25% of all cholesterol in the body is found in the brain [34]. APOE, as a major apolipoprotein produced primarily by astrocytes, plays a pivotal role in maintaining the cholesterol/lipid homeostasis in the brain [2, 35]. Once secreted, APOE binds cholesterol and phospholipids through interactions with the ATP-binding cassette transporter A1 (ABCA1), forming HDL-like particles in ISF and CSF. APOE mediates the interactions of APOE-containing lipoproteins and lipid complexes with cell surface heparin sulfate proteoglycan (HSPG) and cell membrane associated receptors, including the LDL receptor and the LDL receptor-related protein (LRP), promoting cellular uptake and redistribution of cholesterol/lipids, storage, or conversion of cholesterol to oxysterols for clearance through the blood-brain barrier [2, 35]. Importantly, APOE isoforms exhibit differential abilities of binding/transporting cholesterol and phospholipids (APOE2>APOE3>APOE4) [36, 37]. Consequently, APOE4 is poorly lipidated compared with APOE2 and APOE3 [16, 38–41]. Intriguingly, a recent study shows APOE isoform-dependent effects on homeostasis of specific phospholipids, such as phosphoinositol biphosphate (PIP2) [42]. Such differences in lipidation and cholesterol/lipid homeostasis-maintaining capacity of APOE isoforms may account for APOE isoform-specific association with AD [4, 43, 44].

APOE in β-amyloidosis and tau pathology

The role of APOE in β-amyloidosis has been extensively studied and reviewed, in particular on Aβ aggregation and clearance [3, 16, 45]. Recent evidence highlights the role of APOE4 in driving early pathological accumulation/seeding of Aβ in the brain of AD model mice [46, 47]. In addition, a recent study shows experimental evidence that cholesterol plays a catalytic role in Aβ aggregation and that lipid membranes containing cholesterol increases the rate of Aβ aggregation [48]. Thus, the influence on cholesterol trafficking is a possible driving mechanism for the role of APOE in early pathogenesis. Notably, cholesterol-independent effects of APOE on APP/Aβ metabolism were shown recently in embryonic stem cell-derived human neurons [49]. In this study, APOE2, APOE3, and APOE4, either with or without cholesterol, exhibited differential stimulation on APP expression and Aβ production, with a potency rank order of APOE4 > APOE3 > APOE2. Further, it was revealed that APOE exerted those effects through a signal transduction pathway involving the activation of a non-canonical MAP kinase cascade [49].

The gain of toxic effects for APOE4 has been demonstrated recently using induced pluripotent stem cell (iPSC)-derived human neurons. APOE4-expressing neurons exhibited elevated Aβ production, tau phosphorylation, and GABAergic neuron atrophy, when compared to those isogenic neurons expressing APOE3 [50]. Highlighting the unique influence of APOE on different cell types, another similar study evaluated the role of APOE4 in neurons and glial cells. The authors found that transcriptional alterations associated with APOE4 were most commonly related to synaptic function in neurons, lipid metabolism in astrocytes, and immune activity in microglia [51]. While APOE4 neurons released greater levels of Aβ into the culture medium, microglia and astrocytes failed to clear the toxic peptide as effectively as their isogenic APOE3-expressing neurons. The use of iPSCs from human patients is a significant advance in the field of AD research, which offers unique insights into the role of APOE isoforms in human pathology. The use of iPSCs, differentiated to take on peripheral cell phenotypes, in particular vascular cells, may improve our understanding of the role of peripheral APOE in the pathological progression of AD as well.

Furthermore, the Aβ-independent role of APOE in tau pathology and tau-mediated neurodegeneration was recently demonstrated in animal models [52]. Compared with APOE2 and APOE3, APOE4 caused significant elevation of tau levels in the brain of human tau-expressing mice, followed by a marked increase of brain atrophy and neuroinflammation, whereas ApoE-deficient mice were protected from these changes. Further, the presence of the APOE4 allele was associated with more severe regional neurodegeneration in individuals with a sporadic primary tauopathy [52]. Consistent with the role of APOE4 in promoting tau pathology, APOE4-associated phospholipid dysregulation was shown to contribute to the development of tau hyperphosphorylation after traumatic brain injury [53]. In addition, APOE may also affect tau pathology through its effects in the periphery. It has been shown that in young healthy patients carrying the APOE4 allele, circulating peripheral lymphocytes express elevated levels of glycogen synthase kinase 3β (GSK3β) and phosphorylated tau, which were associated with subjective cognitive impairment [54]. In contrast, a recent study reports the association of APOE2, not APOE4, with increased tau pathology in primary tauopathy [55]. It is worth noting that the association of APOE4 or APOE2 with tau pathology may depend on the presence of amyloid pathology [56]. Clearly, more studies are needed to clarify the role of each APOE isoform in regulating tau aggregation under physiological and pathological conditions. Whether peripheral APOE pools participate in these processes also requires further investigation.

APOE in systemic inflammation and neuroinflammation

Compelling evidence indicates that inflammation plays a crucial role in the pathogenesis of AD [5–7] and that APOE modulates immune/inflammatory functions in the periphery and in the CNS [57–60]. Recent studies highlight the impact of both systemic inflammation and neuroinflammation, and their interactions with different APOE isoforms to the progression of AD.

C-reactive protein, a key player of systemic immune response, has been used to investigate the relationship between systemic inflammation and AD. While multiple AD-related genes are associated with the level of CRP [61], the relationship between serum CRP levels and risk of AD has been inconclusive [62–67]. One major limitation in those earlier studies was the measurement of CRP only at one time point. Recently, in a large, longitudinal study of the Framingham Heart Study offspring cohort, multiple measurements of CRP were taken to define a chronic condition of low-grade inflammation at baseline, its interaction with APOE genotype, and its association with the incidence of AD [68]. The results demonstrate that chronic low-grade inflammation interacts with APOE4 to increase the risk of AD and accelerate the onset of AD in a pattern dependent on CRP levels. Importantly, the association between CRP levels and risk of AD for APOE4 carriers became even more significant in the absence of cardiovascular diseases. In addition, while the elevated CRP level was linked to high rates of mortality across all APOE genotypes, an increased risk of AD was observed only in APOE4 carriers. In agreement, another study of a large, ethnically diverse longitudinal cohort showed that among > 100 serum proteins analyzed, only CRP was found to be significantly associated with APOE4 to mediate its contribution to AD risk [69]. In animal models expressing human APOE isoforms, it has also been demonstrated that APOE4, Aβ, and peripheral inflammation interact to induce cerebrovascular damage and cognitive deficits [70]. Furthermore, microRNAs may be involved in mediating the impact of APOE on inflammation. It has been reported that the level of miR146a, a major regulator of innate immune function, is significantly reduced in both the brain and plasma of APOE4 targeted replacement mice, leading to heightened pro-inflammatory signaling centrally and systemically in the absence or presence of Aβ [71]. Together, these findings indicate that APOE4 interacts with chronic systemic inflammation to increase the risk of AD.

In the meantime, much progress has been made lately in understanding the contribution of innate immunity in the brain to the pathogenesis of AD, in particular since the discovery that mutations in the triggering receptor expressed on myeloid cells 2 (TREM2) were associated with increased risk of AD [72, 73]. The role of TREM2 in AD has been covered recently by excellent reviews [74–77]. The focus here is on the APOE-TREM2 interactions and their roles in AD. Following the discovery of association between TREM2 mutations and AD, several studies have identified APOE as a ligand for TREM2 and the APOE-TREM2 interaction modulates phagocytic function of microglia [78–82]. In particular, it has been shown that the APOE-TREM2 pathway drives the dysfunctional microglial phenotypes in neurodegenerative diseases including AD [80, 83]. Using RiboTag translational profiling approach, a recent study discovered that aging, amyloid, and tau pathology converge on APOE pathway in microglia [84]. In addition, plaque-associated APOE origins from microglia and requires TREM2, suggesting microglial APOE contributes to AD pathogenesis [85]. Consistently, deletion of ApoE reduces microglial activation and neuroinflammation in the presence of Aβ or tau pathology [52, 86]. Importantly, the immunomodulatory role of APOE in AD and neurodegeneration may differ depending on the stage of the disease [87]. Regarding different APOE isoforms, a small clinical study showed that the presence of APOE4 allele was necessary to develop AD in TREM2-R47H carriers. However, another study with a larger cohort showed that many TREM2-R47H carriers did develop pathologically confirmed AD in the absence of APOE4 allele. Although an in vitro study found no evidence for APOE isoform-dependent differences in binding to or signaling through TREM2 [82], whether APOE isoforms differentially interact with TREM2 in vivo awaits further investigation. In addition, since astrocytes are the major source of APOE in the brain, it is unclear whether there are any structural and functional differences in APOE from astrocytes and microglia.

While the majority of research on TREM2 has focused on microglia in the brain, it is noteworthy that TREM2 is also expressed on myeloid cells in the periphery, including dendritic cells, mononuclear cells, and tissue macrophages [74, 88]. Recent studies have reported that TREM2 expression is increased in the peripheral blood mononuclear cells in AD patients [89], and in MCI patients who later progress to AD [90]. Whether peripheral pools of APOE and TREM2 interact in the plasma or at the BBB remains to be evaluated, Furthermore, it will be important to understand whether this interaction differs in the context of the periphery and the brain.

APOE in BBB integrity and function

While APOE exists in two distinct pools in the human body, the influence of each pool on the other, and on AD pathogenesis, remains an important area of research. The most obvious place for the differential influences to converge is at the junction of the periphery and the CNS, the BBB. Studies have shown that the majority of brain Aβ (~ 85%) is cleared across the BBB and the remaining fraction is removed by ISF bulk flow along perivascular spaces [91]. The importance of APOE at the BBB is highlighted by the fact that most cells participating in the formation and maintenance of this complex biological barrier, including astrocytes, pericytes, endothelial cells, smooth muscle cells, and neurons (when stressed), express APOE and/or its receptors [16]. Other cells known to interact with the BBB, such as macrophages and microglia, also express APOE.

It is well known that cerebrovascular damage is a hallmark of AD, as some 80% of all AD patients have cerebral deposition of amyloid, known as cerebral amyloid angiopathy (CAA), which is associated with disruption of the BBB [92]. This dysfunction of the cerebral vasculature has recently re-entered the spotlight in the AD field [93–95], and the role of APOE in this process is an area of great interest. Imaging studies show that in cognitively normal APOE4 carriers, there is a widespread and Aβ-independent cerebrovascular dysfunction, indicated by cerebral blood flow (CBF) reduction, impaired cerebrovascular reactivity, and impaired neurovascular coupling [94]. In particular, APOE4 increases the prevalence and severity of CAA [96, 97]. In animal models, early studies indicated that ApoE deficiency leads to damage of the BBB [91, 98–100]. More recently, it has been shown that the influence of APOE on BBB integrity, as well as cerebral blood flow, is isoform-dependent [101–103]. In APOE4-expressing AD mice, endotoxin-induced peripheral inflammation resulted in reduced cerebral vessel coverage, leakiness of the BBB, and elevated CAA [70]. Further, in a bioengineered 3D model of human vessels, it was shown that while circulating APOE (representing that from the peripheral pool) in general facilitates the clearance of Aβ across the synthetic BBB, APOE4 was deficient in this capability compared to APOE2 [104]. Moreover, efforts are underway to understand the role of peripheral APOE in BBB integrity and function in vivo. It was reported that in mice expressing human APOE3 and APOE4 only by the liver in the mouse ApoE-knockout background, peripheral APOE4, but not APOE3, compromises the BBB integrity and cognitive function [105]. Mechanistically, the role of pericytes in maintaining BBB integrity and clearing vascular Aβ has been underscored recently. APOE2 and APOE3, but not APOE4 or lack of APOE, were shown to suppress the pro-inflammatory signaling cyclophilin A-nuclear factor-κB-matrix-metalloproteinase-9 pathway in pericytes to maintain BBB stability [102]. Similarly, it was found that pericytes actively internalize and degrade Aβ at the BBB, which requires APOE and LRP1; APOE3, but not APOE4, restores Aβ clearance by pericytes when the endogenous ApoE gene is silenced [106]. Thus, pericytes function in an APOE/LRP1 dependent and APOE isoform-specific manner. Together, these findings indicate that amyloid, APOE4 status, and peripheral/central inflammation - all known to induce vascular damage in their own right - may act synergistically to wreak havoc on cerebral blood vessels and the BBB.

APOE in Synaptic Plasticity and Cognitive Function

The role of APOE and its isoforms in synaptic and cognitive function is well documented [107, 108]. Earlier studies have shown that there are APOE isoform-dependent changes in hippocampal synaptic function [109–111], and that APOE4-expressing mice exhibit age-dependent cognitive deficits [112, 113], as reported in humans [114, 115]. Recent studies have unraveled some molecular mechanisms underlying these changes associated with APOE4. It was shown that APOE4 selectively impairs APOE receptor trafficking and recycling, leading to reduced glutamate receptor function and synaptic plasticity [116]. Impairment of GABAergic interneurons was identified to cause cognitive deficits in APOE4 mice [113, 117], and enhancing GABA signaling rescues learning and memory function in APOE4 mice [118]. The involvement of GABAergic interneurons in APOE4-induced synaptic dysfunction is further supported by a recent report that APOE4-associated neuronal hyperactivity is caused by reduced responsiveness of excitatory neurons to GABAergic inhibitory inputs [119]. In addition, a recent study demonstrated that APOE4 mice exhibit age-dependent disruption in hippocampal network events critical for memory processes, which depends on GABAergic interneurons [120]. These findings indicate the detrimental effects of APOE4 on synaptic and cognitive function and the importance of the GABAergic system in the maintenance of the excitatory/inhibitory balance in the brain.

The role of peripheral APOE in synaptic/cognitive function has not been extensively studied. As discussed above, APOE4 carriers have low levels of plasma APOE, which may contribute to the increased risk of AD associated with AD. Supporting the connection of peripheral APOE to healthy aging, a recent clinical study of individuals aged 56–105 years showed that centenarians had the highest plasma APOE levels and the lowest frequency of APOE4 allele compared to younger groups [121]. Another study found that peripheral APOE4 was associated with increased loss of grey matter volume in the posterior cingulate, and reduced glucose metabolism in the hippocampus [122]. Furthermore, in cognitively normal E3/E4 carriers, females with higher plasma APOE3 levels performed better on the verbal reasoning (similarities) subtest of the Wechsler Adult Intelligence Scale test. Experimental evidence supporting the importance of plasma APOE came from a recent study, in which expression of APOE only in the periphery improved cognitive performance and partially restored synaptic defects in ApoE-deficient mice [123]. Notably, this study also highlights the importance of APOE in the brain, as synaptic loss and dysfunction still occurred in the brain of ApoE-deficient mice, despite improved memory in the presence of plasma APOE. The role of peripheral APOE in cognitive function is further supported by an elegant recent study, in which ApoE-deficient mice were modified to express either APOE3 or APOEE4 only in the liver; it showed that peripheral APOE3, but not APOE4, enhanced learning and memory performance in the absence of brain APOE [105]. These findings, collectively, indicate that both peripheral and central pools of APOE play important roles in maintaining neurological health.

Emerging therapeutic strategies targeting APOE

As the strongest genetic risk factor for AD, APOE represents the most attractive therapeutic target in AD. However, whether to reduce or increase APOE remains a question of debate, especially with APOE4 exhibiting both loss and gain of functions. In addition, although there is evidence that APOE might be dispensable for cognitive function in the brain, APOE deficiency in the periphery has serious consequences [124–126]. In order for an APOE-targeted treatment to be effective, it will be important to understand the differential and cooperative roles that the peripheral and central pools of APOE play in AD pathogenesis, as well as the impact of different APOE isoforms, as discussed in previous sections. Currently pursued therapeutic strategies targeting APOE are summarized in Table 1.

Table 1.

Currently pursued therapeutic strategies targeting APOE

Agent Class	Target	Proposed Mechanisms	Outcomes	References
Antibodies	APOE	Binds Aβ plaqueassociated APOE, specific APOE isoforms (i.e. APOE4), or conformations (i.e. nonlipidated, aggregated)	Reduces Aβ plaque deposition and microglial activation, and improves memory in mouse models of AD	[127–129]
Nuclear receptor agonists	LXR, RXR	Upregulates APOE and ABCA1 to produce higher levels of APOE, which is more lipidated	Reduces Aβ plaque burden and improves memory in mice of all APOE genotypes; conflicting results in human phase II clinical trial	[130–135]
Apolipoproteinmimetic peptides	APOE, APOA-I, APOJ, and their receptors	Anti-inflammatory, cardioprotective, improves APOE lipidation, mitigates amyloid toxicity	Reduces Aβ plaque burden, protects against neuronal damage, and improves memory in AD mice; safety demonstrated in human phase I clinical trials for some peptides.	[139–151]
Structure correctors	APOE4	Restores the tertiary structure of APOE4 to that of APOE3	Reduces neurotoxicity and Aβ build-up in human stem cells	[50, 154]
Genetic approaches (viral vectors)	APOE2, APOE4	Converting APOE4 to APOE2, or overexpressing APOE2 in the presence of APOE4	Reduces inflammation and BBB damage, improves APOE lipidation, and reduces Aβ plaque deposition in mice; currently being tested in a phase I human clinical trial.	[39, 155]

One exciting approach is passive immunotherapy with APOE-targeted antibodies. Earlier studies showed that antibodies targeting endogenous mouse APOE reduced amyloid accumulation in the brains of mice either before or after the onset of amyloid plaque formation [127, 128]. Recently, it was further shown that antibodies specifically targeting nonlipidated, aggregated human APOE4 inhibited amyloid accumulation in another AD mouse model [129]. Importantly, similar results were obtained with both a central and peripheral delivery of the antibodies, and neither approaches caused significant changes in plasma or brain APOE levels. These findings suggest that the effects of anti-APOE antibodies are not mediated simply by lowering total APOE levels, raising the question of what mechanisms underlying the Aβ-clearing effects of these antibodies. Experimental evidence suggests that microglial-mediated Aβ phagocytosis could be involved in the action of these anti-APOE antibodies [129]. Further studies are needed to unravel the exact mechanisms and to determine any potential adverse effects of this promising approach.

Recognizing the importance of the lipidation state of APOE as well as its overall levels, many groups have focused efforts on drugs targeting the nuclear receptors, e.g., liver X receptor (LXR) and retinoid X receptor (RXR) agonists, as potential therapeutics in AD. These agents have been demonstrated to upregulate ABCA1 and APOE, increase APOE lipidation, and to reduce amyloid pathology and cognitive impairment in mouse models of AD [130]. The most publicized of these agents is bexarotene, owing to its already being a clinically approved agent for the treatment of T-cell lymphoma. Bexarotene was shown to mitigate AD-related pathology and memory deficits in animal models, although some controversy around the effects of this agent on amyloid deposition has arisen [131, 132]. Further complicating the story, a small placebo-controlled phase II clinical trial of bexarotene showed no alteration in amyloid levels in AD patients [133]. However, subgroup analysis determined that APOE4-carriers were holding back the overall analysis, as those carrying APOE2 and APOE3 did in fact have reductions in amyloid burden. This is in contrast to studies in animal models, where bexarotene was shown to mitigate APOE4-driven pathology [134, 135]. In addition, the adverse peripheral effects of bexarotene on triglyceride production and liver health [136–138] are of great concern for long-term use of this class of drugs for prevention or treatment of AD.

Alternatively, several groups have investigated the potential of apolipoprotein/high density lipoprotein (HDL) mimetic peptides for modulating APOE lipidation and function. These HDL mimetics possess anti-inflammatory and anti-oxidative properties, and exert vascular-protective effects [11]. A number of these peptides, derived from or based upon key receptor-binding regions of APOA-I, APOE, or APOJ have been shown to improve cognition, reduce neuronal damage, and to mitigate AD-related pathology in animal models [139–146]. One particular HDL-mimetic peptide, called 4F, has been tested in three clinical trials for cardiovascular disease and was shown to be safe and well tolerated in humans [147–149]. Recent work in our own laboratory with 4F indicates that this unique peptide promotes APOE secretion and lipidation, counteracts Aβ-induced inhibition of APOE secretion and lipidation in primary astrocytes and microglia [150], and reverses APOE4 lipidation deficiency [151].

Another intriguing strategy is based on the observation that differences in APOE protein structure and domain interaction in the nonlipidated or poorly lipidated state underlie the detrimental effects of APOE4, compared with other APOE isoforms [152, 153]. Notably, small-molecule structure-correctors have been designed to restore APOE4 to an APOE3-like tertiary structure, which has been shown to rescue APOE4-associated neuropathology [154]. Recently, this approach was shown to mitigate toxic effect of APOE4 in neurons derived from human iPSCs [50]. Whether these promising molecules affect peripheral and CNS APOE levels and function in vivo awaits further investigation.

Gene therapy is also an important potential therapeutic avenue for targeting APOE. It has been demonstrated that astrocyte-mediated expression of APOE2, delivered intraventricularly via adeno-associated virus (AAV) into the brains of APOE4 mice, improves lipidation and reduces amyloid deposition [39]. This study also demonstrated that injecting APOE4 into APOE4 carriers worsened pathological hallmarks of AD, implying that treatments generally increasing APOE expression may be detrimental in those carrying one or more APOE4 alleles [39]. Importantly, a phase I, open-label clinical trial to test the safety of APOE-targeted gene therapy (NCT03634007) is set to begin recruiting soon. This trial will employ cisternal delivery of an AAVrh.10hAPOE2, serotype rh.10 AAV gene transfer vector expressing human APOE2, into APOE4 homozygotes with positive amyloid PET scans and clinical diagnosis of MCI or AD to determine the maximum tolerated dose, as well as the level and pattern of APOE2 expression in the human brain. This viral vector and route of administration were identified in a previous study to be the safest method for the most widespread delivery of APOE2 in a non-human primate model system [155]. Studies designed to determine the efficacy of this treatment in improving cognitive function in AD will need to wait on the results of this trial. Advances in genetic engineering, such as CRISPR/Cas9, make this an increasingly exciting prospect [51], but much remains to be determined regarding the safety and feasibility of this relatively new technology for prevention or intervention of human disease. Central delivery of any agent is challenging, costly, and potentially dangerous; a fact that is exacerbated by the use of a viral vector, which is likely to induce an inflammatory response as well. An intriguing avenue, supported by many studies discussed in this review, may be to manipulate the APOE genotype in the liver, to influence the make-up of the peripheral pool of APOE.

Limitations of current knowledge

The findings presented here represent significant recent advances in our understanding of APOE biology and its relevance to AD pathogenesis and treatment. Nevertheless, there are limitations to these studies and many knowledge gaps remain. Perhaps the most important limitation herein is our reliance on animal studies, in particular mouse models. While the use of mouse models enables a much wider array of studies than would be possible with human subjects, the relevance to human physiology is not always achieved. As far as lipoprotein metabolism is concerned, one of the major differences between mice and humans is that mice lack cholesteryl ester transfer protein (CETP) [156], which causes HDL being the predominant lipoprotein particles in the circulation of mice. Whereas in humans, the predominant circulating lipoproteins are LDL particles. Thus, mice and humans have very different plasma lipoprotein/cholesterol profiles, consequently different susceptibilities to metabolic and cardiovascular disease. More specifically, human and mouse APOE differ in many respects. Unlike humans, mice only have one form of APOE. Mouse and human APOE only share 70% homology in amino acid sequence [157], and the upstream promoter region of APOE gene shares only 40% homology [158]. Importantly, while both proteins bind the canonical APOE receptors, they also have unique binding partners that are not shared between the homologs. In addition, while APOE is extensively glycosylated in human brain tissue and CSF, mouse brain APOE is scarcely glycosylated [18]. Furthermore, it has been shown that mouse APOE behaves most similarly to human APOE3, as opposed to APOE4, in that it preferentially binds HDL over LDL particles, and does not produce the ‘domain interaction’ thought to underlie some of the detrimental effects of APOE4 [159]. Although human APOE knockin mice have helped addressed some of the primary differences in APOE per se, the inherent lack of CETP in mice remains problematic as a model. Due to the lack of CETP, APOE is necessary for the formation of spherical, lipidated HDL particles in mice; whereas it is not required for genesis of mature HDL particles in humans [160]. Therefore, differences between mouse and human APOE and associated lipoprotein metabolism should be considered when results from mouse models are translated to humans.

Another limitation in our current understanding of APOE is the differential role of therapeutics in the context of APOE genotype. Antibodies targeting APOE2, for instance, may be detrimental, while those targeting APOE4 could produce more protective effects. Similarly, APOE structure corrector molecules would only be hypothesized to provide benefit in APOE4 carriers. On the other hand, Bexarotene and other agents targeting nuclear receptors may be more effective when promoting the expression of APOE2 and APOE3, as opposed to APOE4, as discussed previously. Identifying these differential effects in human carriers of one or more APOE genotype will be critical for the future success of APOE-targeted therapeutics.

Concluding remarks

The current evidence highlights the far-reaching impact of APOE in AD. From peripheral tissues to different cell types in the brain, APOE plays a pivotal role, and APOE4 drives pathology (Fig. 2). With a continued need for effective therapeutics to halt the clinical progression of AD, and setbacks in recent clinical trials, APOE has rapidly risen to the forefront as a druggable target. It will be important to understand how APOE-targeted therapeutics modify potential comorbidities, including cardiovascular disease and systemic inflammation, as well as any potential adverse events associated with systemic administration of these agents. While central pools of APOE drive key pathological hallmarks of AD, including amyloid deposition, tau hyperphosphorylation, and neuroinflammation, peripheral APOE contributes to systemic inflammation, amyloid clearance, BBB integrity, and overall vascular health. In both the periphery and in the brain, APOE4 is a key driver of pathological progression. Many questions remain unanswered as to the direct and indirect impact of peripheral APOE on cognition and AD, and a focus on this area will be vital to continued efforts to combat this horrendous disease.

Fig. 2.

APOE4 in the periphery and in the brain drives the pathological progression to AD.

Highlights.

• APOE exists in two distinct pools, in the periphery and in the brain.

• APOE regulates multiple molecular pathways in addition to cholesterol homeostasis.

• The three APOE isoforms exert differential effects on Alzheimer-related processes.

• Both peripheral and central pools of APOE contribute to Alzheimer’s disease risk.

• Promising APOE-targeted therapeutic strategies have emerged for Alzheimer’s disease.