Genetic and biochemical evidence suggest that apoE plays a crucial role in the pathology of AD [3
]. In the CNS, apoE is the major component of HDL-like particles that deliver lipids to neurons via uptake by apoE receptors such as LRP1 and LDLR. ApoE and the cholesterol efflux pump ABCA1 are LXR-target genes and LXRs are highly expressed in glia cells, which are the primary site of apoE and lipid synthesis in the brain. Both short- and long-term administration of synthetic LXR agonists have been shown to upregulate apoE and ABCA1 levels and to reduce brain amyloid load [17
]. We have shown here for the first time that systemic administration of LXR agonists also leads to a marked and rapid up-regulation of apoE, cholesterol, and Aβ peptides in the CSF of rats, suggesting that this pathway can contribute significantly to the elimination of Aβ from the brain.
In addition to providing cholesterol needed for synaptic plasticity and membrane maintenance, apoE is involved in the elimination of excess cholesterol from the brain and this pathway is upregulated in the neurodegenerative state [35
]. In agreement with this important apoE role, administration of LXR agonist T1317 increased brain cholesterol excretion by almost 3-fold and slowed neurodegeneration in Niemann-Pick type C (NPC1) mutant mice [36
]. Deletion of LXRs in mice results in several brain abnormalities, most notably the lateral ventricles are closed, lined with lipid-laden cells, and deprived of CSF content by 12 months of age [37
]. Deletion of LXRs also significantly increases brain amyloid burden in APP-Tg mice [38
]. Taken together, these studies implicate LXRs in the clearance of excess cholesterol and Aβ from the brain, likely involving transport to the CSF.
Amyloid-binding apoE and apoAI are the most abundant apolipoproteins in the CSF [39
]. However, while CSF apoAI is derived from the plasma, CSF apoE is believed to originate entirely within the CNS [40
]. CSF is produced by the choroid plexus, the ependymal cells lining the ventricles, and the brain parenchyma [41
]. Besides offering support and cushion to elements of the CNS, the CSF also provides an effective pathway for elimination of metabolites from the brain via drainage to the vascular system [31
]. In young and adult rats, CSF is secreted at a rate of approximately 1.3 μl/min and is replaced in its total volume about 10 times per day [42
]. In humans, the total CSF volume is approximately 130 ml and is completely replaced about four times per day [31
]. CSF secretion rate decreases with aging, but the overall CSF volume increases in older individuals due to a reduced rate of turnover [42
]. These age-related changes may translate into reduced Aβ elimination via changes in CSF fluid dynamics. We found that apoE and ABCA1 mRNAs were highly upregulated by LXR agonists in the ventricular region of the mouse brain. In support of our findings, it has been previously shown that treatment with the endogenous LXR ligand 24S
-hydroxycholesterol led to a marked increase in ABCA1 levels and cholesterol release from the apical side of the rat choroid plexus epithelial cell line TR-CSFB3 [44
]. Furthermore, LXR-mediated increase in cholesterol release was greater when apoE3 instead of apoE4 was used as an acceptor [44
]. Taken together with these previous published findings, our results suggest that LXRs play an important role in controlling the excretion of lipid-associated apoE into the CSF and that this mechanism may underlie changes in the ability of the CNS to control amyloid clearance via lipid particle transport out of the brain parenchyma.
The steady-state levels of brain ABCA1 were also rapidly increased with LXR agonist treatment in rats, consistently with the overall changes in apoE and cholesterol seen in the CSF. The changes induced by GW3965 treatment for ABCA1 in the mouse brain were of similar magnitude to that observed in GW3965-treated rats, and about 1.5-fold greater than the levels observed in vehicle treated controls. Correspondingly with LXR agonist treatment, a 6-fold genetic overexpression of ABCA1 in the brain led to a striking decrease of amyloid burden in APP-Tg mice [45
]. Overexpression of ABCA1 was further associated with increased lipidation of astrocyte-secreted apoE, but overall reduced levels of brain apoE, suggesting that highly lipidated apoE particles have an improved clearance rate from the brain parenchyma. In contrast, ABCA1 deletion from APP-Tg mouse lines decreased apoE secretion by astrocytes and brain apoE levels with concomitant increase in amyloid deposition [46
], further implicating apoE and lipid transport as an important route for amyloid clearance in the CNS.
Upon T1317 treatment, the CCF-STTG1 human astrocytoma cell line released cholesterol to exogenously added HEK-apoE3 conditioned media but not to HEK-control vector conditioned media suggesting that apoE is a putative acceptor for ABCA1-mediated cholesterol efflux stimulated by LXR agonists. In addition, the knock-down of ABCA1 prevented cholesterol release to HEK-apoE in the presence of T1317. These results suggest that ABCA1 up-regulation by LXR agonists leading to increased lipidation of brain apoE could modulate apoE's ability to interact with Aβ as well as its catabolism and transport within the CNS.
ApoE is found in association with amyloid extracted from AD brains and in vitro
, apoE forms a SDS-stable complex with Aβ in an apoE isoform dependent-manner [8
]. This differential binding of apoE isoforms to Aβ appears to correlate with their ability to associate with lipids, with apoE4 showing decreased association with lipids and reduced ability to bind Aβ when compared to apoE3 and apoE2 [8
]. The AD-protective apoE2 isoform, which has reduced binding to LDLR and increased association with lipids compared to apoE3, can potentially improve the transport of Aβ out of the brain, likely via the CSF [49
] (Figure ). Because apoE4 appears to have increased clearance by apoE receptors, we hypothesize that the diminished ability of apoE4 to bind Aβ and its faster clearance rate within the brain parenchyma leads to impaired clearance of soluble Aβ, which results in an increased rate of oligomerization and deposition into neuritic plaques (Figure ). The reduced capacity of apoE4 to transport lipids can also lead to increased neuronal susceptibility to toxic Aβ species [50
]. This model is corroborated by previous findings showing decreased CSF apoE in apoE4-KI mice, while human apoE2 expressing mice have increased apoE levels in the CSF when compared to apoE3 [51
]. Furthermore, deletion of LDLR in these mice normalized the level of CSF apoE in the ε3
genotypes to the levels observed in the ε2
genotype, a finding that supports the in vitro
evidence of a faster clearance rate for apoE4 by LDLR within the brain [11
]. By increasing the levels and lipidation of apoE as well as decreasing LDLR levels within the brain, LXR agonists would favor the transport of apoE and amyloid to the CSF subsequently reducing brain amyloid burden, similarly to an "apoE2 effect" (Figure ).
Figure 7 Proposed mechanism by which LXR agonists improve apoE-mediated Aβ clearance from the brain. Based on the differential ability of the apoE isoforms to associate with lipids, Aβ, and LDLR, the model predicts that the "protective" apoE2, (more ...)
It has been previously proposed that the primary mechanism by which LXR agonists decrease Aβ burden is via increased degradation of apoE-lipid by microglia [22
]. This mechanism may involve apoE receptors that are highly expressed in microglia, particularly LDLR. In agreement with the ability of apoE-lipid particles to bind Aβ and promote its clearance, overexpression of LDLR reduced brain apoE levels and Aβ deposition in APP Tg mice [27
]. However, LDLR deletion did not result in increased amyloid deposition in the same APP Tg model [51
]. Unchanged levels of amyloid could have resulted from increased transport of apoE-bound Aβ out of the brain in the absence of LDLR. Nonetheless, due to the high levels of Aβ production in APP-Tg mice, this transport mechanism was likely saturated, which prevented an overall decrease in the brain Aβ burden from occurring.
Even though LDLR is not directly regulated by LXRs, recent findings implicated the LXR-responsive ubiquitin-ligase Idol in the degradation of LDLR and VLDLR in peripheral tissues other than the liver [52
]. Although it did not reach statistical significance, a down-regulation of LDLR and VLDLR levels was observed in the brain of rats treated with LXR agonists for 10 days, in agreement with the Idol mechanism previously described for other tissues. Taken together with these recent findings, our results suggest that LXR agonists not only increased apoE synthesis and lipidation, but also led to reduced levels of apoE receptors in the brain, which could result in further improvement of apoE-bound amyloid clearance from the CNS. This regulatory mechanism is consistent with the pivotal role of LXRs in promoting reverse cholesterol transport from tissues to the liver for excretion.
Our novel findings suggest that LXRs play an important role in the elimination of excess brain cholesterol via secretion of apoE-lipid particles into the CSF. This transport mechanism is also likely to be involved in the elimination of amyloid, particularly of Aβ42. It has been recently shown that low CSF Aβ42 levels are more closely linked to the presence of the ε4
allele of APOE than to clinical status of AD patients, which suggests that Aβ42 levels in the human CSF are highly influenced by apoE genotype [54
]. The APOE4 ε4
genotype is associated with increased brain amyloid load in AD patients and normal individuals, as well as in APP Tg mice [55
]. Recent studies have also shown that apoE4 KI mice have decreased steady-state levels of apoE protein in the CNS when compared to apoE3 KI mice [7
]. Therefore, reduced levels of Aβ42 in the CSF and increased brain amyloid load in apoE4 carriers could be explained in part by the reduced capacity of apoE4 to bind and transport amyloid out of the brain.