Introduction to apolipoprotein E and the cerebrovasculature (CV) (Fig. 1)

APOE4 is the greatest genetic risk factor for Alzheimer’s disease (AD), is associated with cognitive changes during aging and increases risk for other disorders, e.g., Lewy body disease, compared to APOE3. Thus, there is a critical need to dissect pathways modulated by APOE. There are three polymorphic alleles of the human APOE gene, ε2, ε3 and ε4, which encode three isoforms of apolipoprotein E (apoE, 299 amino acids, 34 kDa, reviewed extensively in [58, 67]). Human apoE isoforms differ at residues 112 or 158: apoE2 contains Cys112, Cys158; apoE3 contains Cys112, Arg158; and apoE4 contains Arg112, Arg158. Mouse apoE is structurally and functionally distinct from human apoE and is not the focus of this review. ApoE plays a key role in a number of biological processes in the periphery and the central nervous system (CNS). In the periphery, apoE is important for cholesterol metabolism and APOE4 is associated with hyperlipidemia, hypercholesterolaemia, atherosclerosis and coronary heart disease. In the CNS, apoE modulates multiple mechanistic pathways that collectively affect cognition including cholesterol/lipid homeostasis, synaptic function, glucose metabolism, neurogenesis, mitochondrial function, tau phosphorylation, neuronal atrophy, neuroinflammation, and the metabolism and aggregation of amyloid-β (Aβ). Peripheral apoE is produced by the liver and macrophages but peripheral apoE-containing lipoproteins do not cross the CV, thus apoE is also produced locally within the brain. ApoE is produced by glia (astrocytes and microglia), pericytes, smooth muscle cells, and to a much lesser extent in neurons under certain stress conditions, and potentially brain endothelial cells (BECs). The major apoE receptors are part of the low-density lipoprotein receptor (LDLR) family, which are expressed throughout the CNS. Astrocytes, microglia, neurons and BEC express LDLR, LDLR-related protein 1 (LRP1), very LDL receptor (VLDLR) and LRP8/ApoE receptor 2 (ApoER2) [58]. LDLR and LRP1 are endocytic receptors and the main apoE metabolic receptors, while VLDLR and ApoER2 are primarily signaling receptors. ApoE also binds to heparan sulfate proteoglycans, which facilitate receptor-independent and receptor-dependent apoE uptake. Therefore, APOE-modulated effects on peripheral and CNS pathways, as well as direct apoE signaling in the cells (astrocytes, pericytes, BECs) of the neurovascular unit (NVU), may collectively induce CV dysfunction and cognitive decline. However, research on the role of apoE in CV dysfunction has not advanced as quickly as other apoE-modulated mechanistic pathways. This review will assess the role of APOE in CV dysfunction in aging and neurodegenerative disorders, discuss potential mechanisms and summarize the therapeutic significance of the topic.

Fig. 1
figure 1

Expression of APOE and apoE receptors in the CNS. a Arterioles. b Capillaries that define the blood–brain barrier (BBB)

Full size image

Aging/disease-independent effects (Supplementary Table 1a)

APOE modulates cognitive function during aging

As this article is focused on whether APOE modulates CV function and impacts cognition, an important consideration is the effect of APOE on cognitive function in aging. In young adults, evidence supports that APOE4 imparts behavioral advantages in tasks that require a wider spatial and temporal attentional field [97]. In middle-aged adults there is no clear consensus on whether APOE4 modulates cognitive ability (40–55 years, reviewed in [101]). In older adults (>55 years), APOE4 is associated with cognitive deficits in logical memory, recognition memory and processing speed as well as delayed recall and subjective memory complaints [20, 33, 57, 62, 72]. Proposed hypotheses for APOE-modulated cognition during aging include: APOE4 modulated pathways induce neuronal dysfunction independent of AD; older individuals with APOE4-induced cognitive impairment are in the prodromal AD stage, and antagonistic pleiotropy, i.e., APOE4 exerts beneficial effects early in life, neutral-to-no differences at a mid-age and detrimental effects at old age.

APOE4 synergistically interacts with vascular risk factors to impact cognition

APOE4 could affect cognition later in life through increasing the risk of developing and exacerbating damage caused by cardiovascular risk factors [9]. For example, in APOE4 but not APOE3 carriers, cardiovascular risk factors including hypercholesterolemia, prior cigarette use, diabetes mellitus and hypertension result in longitudinal preclinical memory decline (auditory verbal memory) (mean age 60, 5.6 years follow-up) [21]. Further, the interaction among APOE4, systolic blood pressure and neuropsychological performance was demonstrated in the Framingham Offspring Cohort (mean age 61 years) [141]. APOE4 also potentiates cognitive decline in the absence of pathological hypertension but with increased blood pressure. Normotensive APOE4 carriers with higher systolic blood pressure present with smaller prefrontal volume, slower processing speed and decreased verbal recognition [13]. Although some data conflict [32, 89], overall evidence supports that APOE4 interacts with peripheral cardiovascular risk factors to impact cognition and these factors share common downstream pathogenic properties: atherosclerosis, stroke and BEC dysfunction.

APOE4 disrupts cerebral blood flow

There is an intimate bi-directional association between CBF and neuronal metabolism. CBF ensures sufficient oxygen and nutrient supply, and neurons can secrete factors that influence CBF. APOE modulates CBF when assessed using PET or arterial spin labeling (ASL) MRI. In a small cohort of young college students, APOE4 carriers exhibited lower resting CBF in the left and right inferior temporal gyri and higher CBF in the left insula, right supramarginal gyrus and the inferior occipital gyrus compared to non-carriers [102]. In middle age and older adults (52–81 years) APOE4 is associated with higher CBF as a function of longer sedentary time [148]. These data are consistent with the finding that CBF is elevated for APOE4 carriers in the medial temporal lobes and left parahippocampal and fusiform gyri, the latter of which is positively correlated with verbal memory [133]. Therefore, younger middle-aged adults could display compensatory mechanisms in brain regions at risk for AD, but the data are also compatible with antagonistic pleiotropy. Indeed, CBF is lower in older (50–78 years) compared to younger (20–35 years) APOE4 carriers [36]. Further, with APOE4 there is higher baseline CBF followed by greater CBF decline with age in frontal parietal and temporal cortices in longitudinal analysis (mean age 69.2, interval length 7.8) [121]. In vivo data support the idea that APOE modulates CBF during aging. In APOE4-targeted replacement mice (APOE4-TR mice), which express human apoE4 under the control of the mouse endogenous apoE promoter, there is a reduction in cortical CBF compared to wild type, but not compared to apoE-knockout mice at 18 months of age. Although compared to wild-type mice the changes in functional connectivity were apparent at 12 months in APOE4-TR mice, reduced post-synaptic density levels occurred with perfusion deficits at 18 months [143].

A number of groups have performed fMRI with no clear consensus on the effects of APOE genotype on CBF as assessed by imaging (reviewed in [123]). Evidence for a role of APOE in CBF and CV dysfunction is derived from analysis of blood oxygenation level-dependent functional MRI (BOLD-fMRI). In BOLD-fMRI a signal represents a transient increase in CBF during neural activity, which in turn is dependent on a combination of neuronal activity, neuronal vascular coupling, CBF and general CV function. Young APOE4 carriers (~25 years) display a higher task-related BOLD signal; however, they also demonstrate the lowest CV reactivity when assessed using a CO2 inhalation challenge [114]. It is tempting to speculate that a disrupted CV may underlie the BOLD signal changes in young adults. Middle-aged (50–65) APOE4 carriers exhibit higher resting CBF when assessed by ASL and decreased BOLD and perfusion responses [37]. Collectively, the higher CBF with APOE4 in younger and middle-aged adults may be related to antagonistic pleiotropy or functional compensation due to CV damage and lead to CBF and cognitive impairments in older APOE4 carriers.

APOE2 and APOE4 increase cerebral amyloid angiopathy (CAA)

CAA is the deposition of proteins, including Aβ (particularly Aβ40), in the leptomeningeal medium and small arteries, cortical arterioles and capillaries and is frequently observed with aging. CAA can induce inflammation, fibrinoid necrosis, microaneurysm, microbleeds, transient ischemic attack, hemorrhages and white matter damage. APOE4 is associated with the increased risk of CAA [22, 41] in the occipital lobe [79], neocortex [85] and meninges and correlates with neurofibrillary tangles [85]. Further, APOE4 enhances the amount of Aβ per vessel [4], which may cause CAA-induced hemorrhage. APOE2 carriers are also overrepresented in patients with CAA-related hemorrhage [26, 41, 74–76, 79, 80], and indices of intracerebral hemorrhage severity including hematoma size, functional outcome and mortality are greater in APOE2 carriers [90]. As APOE2 does not influence the severity of CAA, APOE2 is likely a risk factor for hemorrhage of vessels with CAA. ApoE2 binds with lower affinity to LDLR compared to apoE3 and apoE4 in the periphery [131] leading to hyperlipoproteinemia, which could damage the CV and contribute to CAA.

APOE4 induces BBB/blood–CSF barrier dysfunction

Compelling but limited data support that APOE4 induces CV leakiness in humans. In a seminal study, a higher cerebrospinal fluid/plasma albumin quotient (QAlb) was demonstrated in cognitively normal older APOE3/4 carriers (66–85 years) compared to younger APOE3/4 carriers and both younger and older non-APOE4 carriers [47]. In the Rotterdam study, APOE4 was associated with microbleeds regardless of age (mean age 60.3) [86]. APOE4 also modulates the function of the CV as vasoreactivity is lower in younger APOE4 carriers [114] and in older adults [45]. Furthermore, in older adults, APOE4-induced cognitive deficits are amplified by hypertension and with low CO2 vasoreactivity [45]. White matter hyperintensities (WMH) are regarded as indications of CV dysfunction and are associated with changes in white matter integrity. Reports conflict on whether APOE4 increases or decreases the occurrence of WMH or white matter damage in aging [1, 69, 70, 130]. One factor that may underlie these discrepancies is the interaction between APOE and vascular risk factors. Indeed, vascular risk factor-induced deficits in white matter microstructure integrity are exacerbated in APOE4 carriers [128].

Data from APOE-TR mice support APOE4-induced CV dysfunction. Compared to APOE3-TR mice, in APOE4-TR mice CV permeability to dextran is higher at 6 months [12] and permeability to diazepam [2] is greater at 4 and 12 months. However, a recent study failed to find any differences in CV permeability to exogenously administered IgG in 2–3-month-old APOE4-TR or APOE-KO mice compared to wild type [15]. Further, no differences in dextran (3 and 10 kDa) CV permeability were observed between APOE-KO and wild-type mice, but a comparison for the human APOE genotypes was not performed. Global CV leakiness may be mediated by alterations in the BBB. Reduced microvascular length, DNA fragmentation in pericytes and BEC, diminished microvascular coverage by pericytes, reduced CV vascularization and a thinner basement membrane (including lower levels of collagen IV and laminin) are all observed in APOE4-TR mice compared to APOE3-TR mice [2, 12, 48]. Further, despite no changes in glucose transporter expression, 12-month-old APOE4-TR mice exhibit a lower glucose transport into the brain, as well as increased levels of the receptor for advanced glycation end products [2].

Alzheimer’s disease (AD) (Supplementary Table 1b; Fig. 2)

APOE4 increases AD risk

AD, the most common form of dementia, is a rapidly growing worldwide problem facing society and modern science. Subtypes of AD include early-onset/familial AD (FAD, 5 % of all AD cases) and late-onset AD (LOAD, 95 % of all AD cases). The major pathological hallmarks of AD include extracellular plaques of the Aβ peptide and intraneuronal neurofibrillary tangles comprised of hyperphosphorylated tau. Other broad changes in AD include synaptic and neuronal degeneration, lower hippocampal and cortical volume, reduced glucose metabolism, neuroinflammation, impaired insulin signaling, higher soluble Aβ levels and CV dysfunction. Evidence for APOE4-induced AD risk is unequivocal. APOE4 is the greatest genetic risk for LOAD, increasing risk up to 12-fold compared to APOE3, whereas APOE2 reduces risk [58]. APOE4 is also associated with a lower age of AD onset and an increased risk of progression from mild cognitive impairment (MCI) to AD. Mechanistically, APOE4 has been linked to virtually every AD-relevant pathogenic process including Aβ levels, altered Aβ-signaling both directly through Aβ binding and indirectly, and Aβ-independent pathways. There is now little doubt if APOE4 impacts upon the CV and its function in all stages of AD.

Fig. 2
figure 2

CV deficits with APOE4 and Aβ in EFAD mice. CD31 (green) and Aβ (red, using the MOAB-2 antibody) staining in 8-month-old male mice that express APOE4 and overexpress human Aβ (EFAD mice described in [139]). a ×10 magnification, scale bar 100 μm (b–e) ×63 magnification, scale bar 20 μm

Full size image

APOE4 synergistically interacts with vascular risk factors to increase AD risk

APOE4 is not only associated with an increased risk for AD and cardiovascular disease, but the evidence suggests APOE4 and vascular risk factors combine synergistically to exacerbate cognitive decline in AD [43, 55, 73]. For example, neuropathological hallmarks of AD (neurofibrillary tangles, neuritic plaques and CAA) are increased in patients with diabetes who are APOE4 carriers when compared with non-carriers [43]. The combination of hyperglycemia, hyperinsulinemia and insulin resistance observed in type 2 diabetes plus APOE4 exacerbate the development of AD pathology [73].

APOE4 disrupts cerebral blood flow in AD

APOE4 exerts a pronounced effect on CBF in MCI and AD when assessed by ASL-MRI, or single-photon emission computer tomography (SPECT). A number of groups have demonstrated higher CBF in MCI patients with APOE4. CBF is higher in the medial temporal lobes, parahippocampal gyrus, cingulate gyrus, posterior cingulate gyrus and lingual gyrus of patients who are APOE4 positive and have MCI [10, 60, 134]. The effect of APOE4 on CBF in MCI is brain region specific. In the left parahippocampal/fusiform gyrus, CBF is higher with APOE4 in non-demented controls and lower in APOE4 MCI patients, whereas the opposite pattern is observed in frontal regions [134]. The higher CBF in MCI patients who are APOE4 positive may be indicative of compensatory mechanisms in response to stress, or of an ongoing pathogenic response acting on the vasculature (e.g., inflammation, neuronal activity). Although CBF is elevated in posterior brain regions with one risk factor such as APOE4 or MCI, the presence of both results in decreased CBF and a greater likelihood of conversion to dementia [134]. Thus, in the presence of multiple risk factors declining posterior hippocampal function may result in higher CBF to other brain regions as a compensatory mechanism. In AD, APOE4 is associated with cerebral hypoperfusion, including the occipital lobes, middle temporal gyrus, inferior frontal gyrus, anterior cingulate gyrus, claustrum, insula and caudate [51, 65, 66], as well as a greater spread of CBF reductions from the parietotemporal to the frontal area [99]. However, there are also reports of increased CBF asymmetry in APOE4 non-carriers [125], no APOE4-dependent effects in AD patients [96], and a counter argument that APOE4 promotes neuronal dysfunction rather than CV changes. However, the data are most consistent with disrupted CBF with APOE4 in AD.

APOE4 increases capillary CAA in AD

CAA in cortical and leptomeningeal arteries/arterioles of AD patients follows the order; APOE4/4 > APOE3/4 > APOE3/3 [6, 25, 35, 87, 92, 93, 104], and APOE4 increases CAA in the occipital lobes [122]. For example, in one study the prevalence of severe CAA from highest to lowest in AD was 73.4 % for APOE4/4 carriers, 46 % in APOE3/4 carriers and 24.2 % in APOE3/3 carriers [92]. Further, CAA in APOE4 AD patients is more severe [87] and associated with a longer onset period of cognitive decline to death, lower diffuse plaque score, cortical microinfarcts, leukoencephalopathy, enlarged perivascular spaces in the white matter, subcortical and lobar and intracerebral hemorrhages, thrombosis and fibrinogen deposition [6, 25, 35, 54, 87, 92, 93, 104]. APOE4 also leads to substantial CAA compared to APOE3 and also a higher Aβ40/42 ratio in mice that express FAD mutations (FAD-Tg mice) [38]. However, a few studies have reported a lack of association between APOE4 and CAA in AD [83, 112], no link for APOE4 modulated CAA and hemorrhage [25], that CAA in general rather than APOE4 is important for dementia [117] and an association between CAA and lower cognition proximal to death in non-APOE4 carriers [16]. Thus, there is confusion of whether CAA in APOE4 AD patients is a major contributor to overall dementia. One question that remains is how the APOE genotypes correlate with the types of vessels affected by CAA in AD. A body of data supports that the CAA in APOE4 carriers affects cortical capillaries [3, 120, 140]. For example, Thal et al. [120] observed an odds ratio of 4.751 for capillary CAA in APOE4 AD patients. CAA is also linked to amyloid-related imaging abnormalities (ARIA) after passive and active immunization strategies targeting Aβ in AD [98]. ARIA is characterized by vasogenic edema and cortical hemorrhages, is more common in APOE4-AD patients after immunization with antibodies for Aβ and has been linked to CAA [98, 136]. The higher levels of CAA with APOE4 may reflect detrimental changes in the CV with immunization that include splitting of vessel wall, and/or the removal of Aβ from the vessel wall [98]. In contrast to APOE4, there is a negative association of CAA for APOE2 [35]. The evidence that APOE2 increases CAA and ICH in aging, but not in AD, is potentially due to the protection afforded by APOE2 for AD risk through effects on other AD-relevant pathways, e.g., apoE lipidation and inflammation.

APOE4 induces BBB dysfunction in AD

Initial evidence that APOE modulates the BBB (capillaries) in AD was observed by Salloway et al. [100], who demonstrated a thinning of the basement membrane in APOE4/4 compared to APOE3/3 AD patients. When assessed using quantification of agrin (basement membrane protein) staining, the capillary basement membrane area was smaller in APOE4/4 AD patients compared to APOE3/3 AD patients in the prefrontal cortex [100] and APOE3/4 AD patients trended (non-significant) to lay between APOE3/3 and APOE4/4 AD patients. Basement membrane disruptions may be an indication that the BBB is degenerating with APOE4 leading to increased leakiness. Indeed, levels of the plasma protein prothrombin in the prefrontal cortex are higher in APOE4/4 AD patients than APOE3/3 AD patients (significance at the 10 % level) [146]. Further, IgG and fibrin extravasation surrounding microvessels follows the order APOE4AD > APOE3AD > controls in the frontal cortex [47]. In contrast, no APOE genotype-specific effects were observed on the QAlb in AD patients, although there was evidence for a protective effect of APOE3 on BBB leakiness in Creutzfeldt-Jacob disease [59]. There is evidence that APOE modulates microbleeds in AD. In AD, microbleeds are associated with male sex, higher blood pressure, lower CSF Aβ42 and APOE4 [14]. In contrast to the male bias in humans, microbleeds are higher in female mice that express APOE4 and Aβ compared to male mice [18]. Perhaps there are other stressors synergistically interacting with APOE4 in males with higher microbleeds. APOE4 is associated with increased WMH in the parietal lobe compared to non-carriers [17], however, data conflict on whether APOE4 increases WMH risk in AD [17, 49, 61, 78]. Potential confounding factors are that APOE4 and vascular risk factors interact to induce WMH, and that WMH are brain region specific. Our preliminary data also demonstrate that there are CV deficits in mice that express APOE4 and overexpress Aβ (E4FAD mice described in [139] Fig. 2). In 8-month-old male E4FAD mice we have observed characteristics of CV dysfunction including vessels surrounding extracellular Aβ (Fig. 2b), CAA (Fig. 2c, d), the absence of vessels in areas of high Aβ, and vessel degeneration (Fig. 2e). Additional data that apoE modulates the CV include the negative association of APOE-positive capillaries and extracellular Aβ plaques, the positive correlation of APOE and expression of angiogenesis/vasoactive mediators (VEGF, eNOS) in AD brain capillaries [88], and the link between apoE levels and CV leakiness in AD patients with small vessel disease [124]. Collectively these data support that APOE4 increases CV leakiness in AD patients compared to APOE3.

APOE-modulated CV dysfunction in other neurodegenerative conditions

Data suggest that rather than representing a risk factor, APOE4 exacerbates the progression, is inferior or hinders repair mechanisms in vascular dementia (VaD), multiple sclerosis (MS), stroke and traumatic brain injury (TBI), compared to APOE3. For example, APOE4 is linked to greater cognitive impairment in VaD, higher disease progression/severity in MS [107], a detrimental progression or outcome after intracerebral hemorrhage, subarachnoid hemorrhage, stroke [42] and after TBI [28]. Although the CV is important for these conditions, surprisingly the role of APOE-modulated CV dysfunction has not been dissected. Cleaved apoE is considered toxic, but this is also evidence of an ongoing apoE-dependent process. In VaD, cleaved amino-terminal fragments of apoE were identified within blood vessels in an APOE genotype-dependent manner: APOE4/4 > APOE3/4 > APOE3/3 [94]. For MS, in APOE knockout mice subjected to experimental autoimmune encephalitis, a murine model of MS, there are lower levels of proteins in the tight junctions and higher levels of matrix metalloproteinases, suggesting a role for APOE in the progression of disease, but the exact role is still unclear [44, 144]. In TBI, possession of APOE4 promotes CAA, possibly by disruption of the vascular clearance mechanisms for Aβ [145]. Together, these results suggest that it is important to determine whether apoE4 has a detrimental or inferior effect upon cellular and tissue repair mechanisms in the CNS, compared to the other apoE isoforms.

Potential mechanistic pathways underlying APOE-modulated CV dysfunction (Supplementary Table 1; Fig. 3)

Genetic, environmental, physiological and lifestyle risk factors will collectively induce changes in a multitude of biological pathways that determine whether an individual will experience impaired cognition with aging, or begins to progress towards a neurodegenerative disease. Further, the overlapping nature for many of the pathways can result in comorbidities. Accumulating evidence supports that APOE4 modulates a multitude of mechanistic pathways which can affect cognition. However, APOE4, like diabetes, hypertension and aging is a risk factor not a cause. It would be naïve to propose that for every individual, a single risk factor or pathway is the cause of cognitive impairments, especially in the context of chronic conditions, e.g., Aβ causes AD. Similarly, CV deficits are not disease specific and, therefore, are unlikely to be relevant to primary causation. However, multiple risk factor-modulated pathways acting in temporal-specific manner will impact cognition. Within this complexity, we propose that APOE4 induces CV dysfunction through direct signaling at the NVU, and indirectly via modulation of peripheral and central pathways, contributing to cognitive decline in aging and neurodegenerative disorders. Further, that APOE4 predisposes the CV to damage by, and exacerbates the effects of, additional risk factors (e.g., sex, hypertension, diabetes).

Fig. 3
figure 3

Pathways of APOE-modulated neurovascular dysfunction. APOE4 imparts negative effects on a multitude of peripheral and CNS pathways that may contribute to cerebrovascular dysfunction and lead to cognitive impairment. Brain endothelial cells may act as the primary effector interface for these pathways

Full size image

ApoE4 can directly induce detrimental signaling at the NVU

The question of how the apoE isoforms differentially modulate a diverse range of biological processes to affect cognition is a major focus of a number of groups. ApoE produced by the periphery can signal to BEC and cells of the NVU express apoE and the apoE receptors. The structure and receptor binding properties of apoE could exert a strong influence on the CV. Overall, apoE isoform-specific differences are thought to affect the folding, including the receptor binding domain (N terminus) and the lipid binding domain (C terminus), and stability of apoE (reviewed in [58]). In the CNS differences between the apoE isoforms will affect the structure of lipoproteins, and the majority, if not all, of the apoE in the interstitial fluid of the brain is present as a lipoprotein. One proposed hypothesis is that structural differences among the apoE isoforms in the brain results in lower apoE4 lipidation, lower apoE4-containing lipoprotein stability, the production of toxic apoE4 fragments and lower apoE4 levels, which in turn causes impaired apoE4 cellular recycling and altered receptor activity (reviewed in [116]). Specific for the CV, apoE modulates astrocyte–pericyte–BEC interactions [12, 23, 127, 135]. ApoE4 but not apoE3 (from astrocytes) signaling to LRP1 in pericytes is impaired, resulting in higher activation of the cyclophilin A (cypA)-NFkB-matrix metalloproteinase 9 (MMP9) pathway in pericytes [12, 47]. The higher MMP9 levels with APOE4 leads to CV dysfunction by basement membrane degradation and impaired BEC function. Lipidated apoE is also produced by pericytes in vitro [23] and apoE knockdown in isolated human pericytes accelerates pericyte mobility, which is suppressed by supplementing apoE3, but not apoE4 into the media. Importantly, apoE isoform-dependent modulation of pericyte mobility is mediated by a pathway involving LRP1 and RhoA. Often overlooked is the potential that apoE from the CNS activates apoE receptors in BECs to modulate CV function. Indeed, astrocyte-derived apoE signals via LRP1 on BECs in vitro, an effect greater with apoE3 than apoE4, resulting in lower occludin phosphorylation with apoE4 [81]. Further to agonist activity, the accumulation of toxic apoE4 fragments inside of BEC as observed in VaD could induce CV dysfunction. Therefore, apoE4 receptor signaling cascades may be blunted compared to apoE3 in astrocytes, pericytes and BECs resulting in a disrupted CV.

APOE-modulated neuroinflammation contributes to BBB dysfunction

As reviewed extensively elsewhere [115], APOE4 is increasingly recognized as modulating glial-mediated neuroinflammation in aging and AD. APOE4 is associated with a detrimental response to Aβ-independent neuroinflammation (such as LPS or as occurs in aging) and Aβ-induced inflammation. However, the complexity of chronic neuroinflammation, including multiple detrimental and beneficial effects occurring in a temporal and cell-specific manner, has resulted in conflicting functional data for virtually every inflammatory mediator. Potentially, APOE4 induces a detrimental neuroinflammatory phenotype characterized by higher pro-inflammatory and lower anti-inflammatory cytokines [115]. Importantly, many mediators described as inflammatory exert a strong influence of the CV, including BEC function (e.g., TNFα), angiogenesis (e.g., VEGF, EGF) and the basement membrane (MMPs, TIMPs). In addition, an overlooked topic is how BEC themselves produce inflammatory mediators in response to stress. Mechanistically, the effects on glial inflammation may be mediated in part by direct apoE signaling as described above. Thus, apoE through its effects on soluble mediator release from inflammatory cells within the brain may cause and also prime the CV to damage in response to a subsequent hit from the periphery.

Aβ clearance is slower with APOE4 at the CV

Aβ, in insoluble or soluble form, is considered a key mediator of AD progression and can induce direct (BEC) and indirect (via pathways such as inflammation) CV dysfunction. Therefore, APOE could directly modulate Aβ signaling to affect the CV. For example, Aβ-induced detrimental signaling in glia and BEC may be amplified in the presence of apoE4. In addition, Aβ clearance rates are slower with apoE4 at the BBB [7, 30] and via perivascular drainage [24, 48]. Higher levels of Aβ with APOE4 will result in amplified detrimental pathways in the CNS to cause CV dysfunction. After Aβ42 injections in vivo, Aβ clearance into the plasma follows the pattern APOE3 > APOE4 > APOE-KO/wt [8]. At the BBB, when Aβ binds apoE2 or apoE3 (apoE/Aβ complex) clearance is mediated via LRP1 and LDLR, whereas Aβ-bound apoE4 is cleared at a significantly slower rate via VLDLR [30]. ApoE and Aβ also co-localize in the perivascular elimination pathways of AD transgenic mice as well as in control and AD patients [95]. Importantly, after ICV injections of Aβ into APOE-TR mice, Aβ40 aggregates in the perivascular drainage pathway with APOE4 but not APOE3 [48], evidence that perivascular elimination is slower with APOE4. Further, following Aβ immunotherapy plaque-associated apoE is reduced and CV wall-associated apoE increased [98]. This immunotherapy-mediated translocation of apoE from plaques to the CV mirrors the changes in Aβ and is consistent with a proposed role for apoE as an important transporter of Aβ in the brain. Mechanistically, apoE could affect Aβ clearance through direct binding as an apoE/Aβ complex and an important functional distinction is whether the apoE/Aβ complex is part of a plaque, present in the basement membrane or soluble in the ISF. ApoE when part of a plaque or as part of an initial seed for oAβ may drive Aβ pathology and inflammation and also transport Aβ to the CV, apoE-Aβ interactions at the basement membrane are likely involved in perivascular drainage, and soluble apoE/Aβ complex may prevent oAβ formation and promote Aβ clearance (contrasting in vitro data exist [8]).

APOE4 could prime the CV to damage by increasing the risk of developing and potentiating peripheral stress

A recurrent theme is the potentiation of cognitive decline with APOE4 and peripheral vascular risk factors in aging and AD. APOE4 can increase the risk of developing cardiovascular disease contributing to CV dysfunction, e.g., hypertension increases the risk of BBB disturbances and atherosclerosis. Mechanistically, the N and C terminus of apoE4 interact to a greater extent than apoE3, which directs apoE4 to very low density lipoproteins and increases the incidence of atherosclerosis. The apoE4-induced alterations in lipid metabolism can also modulate CV function. There are higher serum cholesterol, phospholipid and triglyceride levels in APOE4 carriers. Lipolysis products from lipoproteins rich in triglycerides, specifically triacylglycerol-rich lipoproteins (TGRL), alter the structure of apoE4, but not apoE3 [118]. These conformational changes influence the way in which apoE4 binds to BECs [5], which in turn may disrupt BBB permeability, allowing for influx of damaging TGRL lipolysis products. Further, many of the peripheral risk factors have a strong inflammatory component, which will directly affect BEC at the CV. One potential mechanism is that CV active inflammatory mediators in the periphery are modulated by APOE, analogous to neuroinflammation. For example, healthy volunteers injected with lipopolysaccharide exhibit hyperthermia and increases in plasma levels of pro-inflammatory markers (TNFα, IL-6) that are higher in APOE3/4 carriers compared to APOE3/3 carriers [39].

Therapeutic significance (Fig. 4)

Directly targeting apoE or the apoE receptors

Directly targeting the structural and functional deficits of apoE4 may ameliorate detrimental changes that cause CV dysfunction (reviewed in [67]. These strategies may also become relevant for apoE3 in certain conditions. If apoE4 is considered detrimental (toxic gain of function) then lowering apoE4 levels using genetic or antibody-based approaches is optimal. If apoE4 is less functional (loss of positive function) increasing apoE4 lipidation by promoting the activity or levels (e.g., via nuclear receptor agonists) of the primary lipid efflux transporter in the CNS, the ATP-binding cassette transporter A1, or correcting the structure of apoE4 becomes the focus. Both approaches show promise in vivo for AD-relevant pathways. Importantly, nuclear receptor agonists may target BBB dysfunction in AD [149], and are protective against CV dysfunction caused by viral infections [53]. However, both strategies have drawbacks including toxicity and lack of specificity for nuclear receptor agonists, or the necessity of accessing apoE before becoming part of a lipoprotein for structural correctors. The development of more advanced genetic strategies could also lead to overexpression of APOE2 in APOE4 carriers [52]. Directly targeting the LDLR family is an alternative way to circumvent a loss of positive function for apoE4 signaling at the neurovascular unit. ApoE mimetic peptides, which are based on the receptor binding region of apoE, reduce edema after focal brain ischemia and lower CV dysfunction after TBI in vivo [19, 63].

Fig. 4
figure 4

Therapeutic targets for APOE-modulated cerebrovascular dysfunction

Full size image

Targeting inflammation at the NVU and the basement membrane

Targeting either the signaling pathways or the soluble mediators produced by APOE-modulated activated glia (astrocytes and microglia) and pericytes may ameliorate CV dysfunction, or prevent the risk with a subsequent additional hit. For signaling in glia and astrocytes, antagonists/inhibitors of TLR4, p38a, cyclooxygenase 2 (COX2) and nuclear receptor agonists suppress pro-inflammatory cytokines and increase anti-inflammatory cytokines in various in vivo and in vitro models of neurodegenerative conditions. For pericytes, targeting the LRP1-rhoA or cypA-NFkB-MMP9 pathway may prove efficacious. In terms of mediators, likely targets include TNFα, and nitric oxide, a vasoactive mediator proposed as a target for CAA-induced CNS damage. Additional targets include raising levels of tissue inhibitors for metalloproteinases to prevent basement membrane disruption, or targeting soluble mediators that induce changes in angiogenesis and BEC function. Surprisingly, BEC pathways are frequently overlooked as a direct for APOE4-induced CV dysfunction, e.g., reactive oxygen species, COX, mitogen active protein kinase and angiogenic signaling. The main advantage of targeting BECs directly is that therapeutics do not have to cross the BBB into the brain to be active.

Targeting peripheral pathways

Therapeutic or lifestyle interventions for the peripheral factors that synergistically combine with APOE4 to cause CV dysfunction may provide a great benefit for neurodegenerative disorders. For example anti-hypertensive drugs, particularly angiotensin receptor blockers, have shown promise in AD but the effects of these drugs correlated with APOE genotype are unknown. For hypercholesterolemia statins have been tested in clinical trials for the treatment in AD with disappointing results. The biological mechanisms behind why hypercholesterolemia increases AD risk are still unclear, but elevated cholesterol levels can disrupt the integrity of the BBB. Thus, it is proposed that BBB penetrant lipophilic statins disrupt cholesterol synthesis in the brain resulting in detrimental effects on cognition whereas less-BBB penetrant hydrophilic statins may be promising [27]. Targeting peripheral inflammation can be achieved using the same approaches as for neuroinflammation and include nonsteroidal anti-inflammatory drugs (NSAIDs); however, results from the clinic are not as promising as observed in epidemiological studies. Potentially, therapeutics that target hypertension, hypercholesterolemia and peripheral inflammation will prove efficacious for cognitive decline in APOE4 carriers with CV dysfunction. There is growing evidence supporting the adoption of lifestyle changes to improve cognition, which partially act through modulating the CV. ApoE is also necessary for the prevention of age-induced CV dysfunction by exercise in vivo [111]. With professional guidance and assistance most individuals, particularly APOE4 carriers, would be able to curtail damaging habits that put them at risk of cognitive damage caused by CV dysfunction.

Conclusion

ApoE4 exerts a strong influence of CV dysfunction, an effect exacerbated by additional risk factors. Identifying APOE-modulated CV-specific mechanistic pathways via basic and preclinical therapeutic research may ultimately lead to prevention and treatment options to improve cognition in aging and neurodegenerative disorders.