The Low Density Lipoprotein Receptor Regulates the Level of Central Nervous System Human and Murine Apolipoprotein E but Does Not Modify Amyloid Plaque Pathology in PDAPP Mice*

Apolipoprotein E (apoE), a chaperone for the amyloid (cid:1) (A (cid:1) ) peptide, regulates the deposition and structure of A (cid:1) that deposits in the brain in Alzheimer disease (AD). The primary apoE receptor that regulates levels of apoE in the brain is unknown. We report that the low density lipoprotein receptor (LDLR) regulates the cellular uptake and central nervous system levels of astrocyte-de-rived apoE. Cells lacking LDLR were unable to appre-ciably endocytose astrocyte-secreted apoE-containing lipoprotein particles. Moreover, cells overexpressing LDLR showed a dramatic increase in apoE endocytosis and degradation. We also found that LDLR knock-out ( Ldlr (cid:2) / (cid:2) onto an we did not a significant change in brain A (cid:1) before or after the onset of A (cid:1) deposition. Inter-estingly, human APOE3 or APOE4 (but not APOE2 ) knock-in mice bred on an Ldlr (cid:2) / (cid:2) background a 210% and 380% increase, respectively, in the level of apoE in cerebrospinal fluid. These results demonstrate that central nervous system levels both are A or it remains a possibility in

key pathological hallmarks of AD is the deposition of the 39 -43-amino acid amyloid ␤ (A␤) peptide in the form of both diffuse (thioflavine-S/Congo red-negative) and fibrillar (thioflavine-S/Congo red-positive) plaques. An abundance of data suggests that conversion of A␤ from soluble to insoluble forms is an early event in the pathogenesis of AD. The A␤ peptide is generated from cleavage of the larger amyloid precursor protein (APP) with the predominant species being A␤ 40 and ,to a lesser extent, A␤ 42 (1). Although accounting for Ͻ1% of all cases, early-onset, autosomal-dominant forms of AD have been identified that share the common feature of an overall increase in A␤ levels or a relative elevation in the more fibrillogenic A␤ 42 throughout life, ultimately resulting in early A␤ deposition and plaque formation. Identification of these familial AD cases has led to the generation of several APP transgenic mouse models that recapitulate many aspects of A␤ deposition and associated pathology (2).
In 1993, the ⑀4 allele of apoE was found to be a genetic risk factor for the most common form of AD (late-onset AD) as well as for cerebral amyloid angiopathy, whereas the ⑀2 allele was shown to be protective (3). Abundant data suggest that apoE is linked to AD and cerebral amyloid angiopathy due to its ability to act as an A␤ chaperone (4) and influence A␤ metabolism. By acting as an A␤-binding molecule, apoE influences the amount of A␤ deposition and the conformation in which A␤ aggregates, as well as A␤ clearance and toxicity in vivo (5)(6)(7)(8). Furthermore, the level of apoE in the brain directly influences all of the aforementioned processes (5)(6)(7)(8). For example, APP transgenic mice lacking apoE develop less A␤ deposition and virtually no fibrillar A␤ deposits (5,9,10). Thus, understanding the cellular events and receptors that regulate apoE levels in brain may give important insights into AD pathogenesis.
ApoE is expressed at high levels in the liver and the central nervous system (CNS), where it is present in lipoprotein particles. Furthermore, apoE in the CNS is derived from within the CNS and not the plasma (11). However, in the CNS, apoEcontaining lipoproteins differ from those found in the periphery in both the amount of sialation and, perhaps more importantly, the type of lipoprotein particle it associates with (12). In the plasma, apoE is in both very low density lipoproteins (VLDLs) and a subset of high density lipoproteins that also contain other apoproteins. In contrast, CNS apoE is in high density lipoprotein-like lipoproteins that are secreted primarily by astrocytes and, when released by cells, contain apoE as their sole apolipoprotein constituent (12,13). Despite this information, whether specific apoE receptors regulate the level of apoE in the brain or mediate apoE-dependent clearance of apoE-binding proteins is unknown. Nearly 20 years ago, Brown and Goldstein (14) elucidated the role of the low density lipoprotein receptor (LDLR) in receptor-mediated endocytosis of lipoproteins in plasma, including apoE. Subsequently, other members of the LDLR family have been identified. Many are expressed in brain, and they have been shown in vitro to bind and endocytose lipid-rich lipoproteins that contain apoE (15). To determine whether LDLR family members influence the level of CNS apoE and A␤ metabolism, an AD mouse model was recently bred to mice lacking the 39-kDa receptor-associated protein (RAP) (16). An increase in A␤ aggregation and deposition was found in these mice. RAP is an endoplasmic reticulum chaperone protein known to influence the folding and level of all LDLR family members, and in this study, levels of both LDLR-related protein (LRP) and LDLR were examined and found to be reduced in the brain of mice lacking RAP (16). Additionally, overexpression of a functional LRP mini-receptor in the PDAPP model led to an increase in the levels of soluble A␤ at advanced stages of deposition, but with no detectable change in A␤ deposition measured histologically (17). Thus, whether other LDLR family members are responsible for brain apoE metabolism and whether they play a role in regulating A␤ levels in vivo remains unclear. Herein, we investigate the role of LDLR in regulation of CNS-derived apoE as well as on brain A␤ levels in a mouse model of AD.

MATERIALS AND METHODS
ApoE Labeling and in Vitro Assays-The immunoaffinity purification of astrocyte-secreted apoE-containing lipoprotein particles has been described previously (18). For apoE degradation studies, astrocyte-secreted apoE3-containing lipoprotein particles were labeled with 125 I using an IODO-GEN kit (Pierce). Chinese hamster ovary cells that lack endogenous LRP (LRP-null) were used to stably express LDLR, an LRP mini-receptor, apoER2, and the VLDL receptor (19). Degradation of 125 I-labeled apoE3 was measured as radioactivity soluble in 20% trichloroacetic acid in the cell culture supernatant as previously described (19). 1 M unlabeled RAP was used to determine specificity, a concentration that inhibits all LDLR family members. For apoE endocytosis assays, apoE lipoprotein particles were labeled with the fluorescent hydrocarbon probe DiI as previously described (20). Wild-type mouse embryonic fibroblast cells or mouse embryonic fibroblast cells lacking LDLR, LRP, or both receptors were used for the DiI studies (21,22). DiI-labeled apoE was applied for 1 h at room temperature, washed with PBS, and fixed with 4% paraformaldehyde.
Animal and Tissue Preparation-For analysis of tissue and plasma cholesterol and CSF apoE, 3-month-old Ldlr ϩ/ϩ and Ldlr Ϫ/Ϫ mice, both on an identical C57BL/6J background, were used (23). For analysis of human apoE isoforms, Ldlr Ϫ/Ϫ mice were first bred to wild-type SJL mice to generate Ldlr ϩ/Ϫ mice on a mixed C57BL/6J ϫ SJL background. These mice were then bred to human APOE-expressing mice (24) that have been maintained on the same C57BL/6J ϫ SJL background such that all mice analyzed for human apoE experiments were on an identical background. For A␤ analysis, we utilized the PDAPP mice that overexpress human APP 751 with a familial AD mutation at position 717 (APP V717F ) under control of the neuron-specific plateletderived growth factor promoter (25). We crossed PDAPP ϩ/Ϫ , Ldlr ϩ/Ϫ mice with Ldlr ϩ/Ϫ mice to generate PDAPP ϩ/Ϫ , Ldlr ϩ/ϩ and PDAPP ϩ/Ϫ , Ldlr Ϫ/Ϫ mice (littermates) on the same mixed genetic background as described previously (7). Hereafter, all mice referred to as PDAPP in this study are hemizygous for the PDAPP transgene (PDAPP ϩ/Ϫ ). Animals were anesthetized with pentobarbital (150 mg/ kg, intraperitoneal) and perfused transcardially with 0.1 M PBS containing heparin (3 units/ml), pH 7.4. One hemibrain was immersionfixed in PBS containing 4% paraformaldehyde overnight at 4°C. After fixation, the brain was cryoprotected in PBS containing 30% sucrose at 4°C. Brain regions from the other hemisphere were dissected and frozen on powdered dry ice before analysis. Complete protease inhibi-tors (Roche Applied Science) were used where indicated. For PBSsoluble A␤ analysis, tissue was first Dounce homogenized in ice-cold PBS with protease inhibitors and immediately spun for 5 min at 20,000 ϫ g. For carbonate-soluble A␤ analysis, the PBS-insoluble pellet was then homogenized in ice-cold 100 mM carbonate, 50 mM NaCl, pH 11.5, with protease inhibitors and immediately spun for 10 min at 20,000 ϫ g. For carbonate-insoluble A␤ analysis, the carbonate-insoluble pellet was lysed for 3 h with rotation in 5 M guanidine, 50 mM Tris with protease inhibitors. All experimental protocols were approved by the animal studies committee at Washington University.
Biochemical and Histological Analysis-A␤ 40 and A␤ 42 were quantified by enzyme-linked immunosorbent assay as described previously (26). A␤ deposition was assessed by immunostaining (3D6 antibody recognizing amino acids 1-5) and stereological analysis as previously described (7). ApoE was measured in cortical tissue that was Dounce homogenized briefly in PBS with protease inhibitors and spun immediately at 20,000 ϫ g for 5 min. Murine apoE was quantified using a sandwich enzyme-linked immunosorbent assay with pooled C57BL/6J plasma serving as a standard for apoE set at 68 g/ml (27) with a sensitivity of Ͻ5 ng/ml. Human apoE was quantified using the same enzyme-linked immunosorbent assay protocol, except that the standard used was purified apoE isolated from human ␤-VLDL (BioDesign, Sako, ME). The 96-well microtiter plates were coated overnight at 4°C with 20 g/ml monoclonal anti-apoE antibody WUE4 (28). All washes were performed 8 times/well with PBS in a standard microplate washer. Plates were washed and blocked with 1% dry milk in PBS for 60 min at 37°C. Plates were washed, and samples and standards were loaded in 0.5% bovine serum albumin in PBS-0.025% Tween 20 containing protease inhibitors and incubated for 90 min at 37°C. Plates were washed and incubated with goat anti-apoE antiserum (Calbiochem) at 1:500 dilution in 0.5% bovine serum albumin in PBS-0.025% Tween 20 for 90 min at 37°C. Plates were washed, and horseradish peroxidase-conjugated horse anti-goat antibody at 1:5000 dilution (Vector Laboratories, Burlingame, CA) was incubated for 90 min at room temperature. Plates were washed, tetramethylbenzidine substrate (Sigma) was added, and absorbance was monitored at 650 nm. The Amplex Red kit (Molecular Probes, Eugene, OR) was used to measure total cholesterol content.
Western Blotting-Tissue samples were homogenized in radioimmune precipitation assay buffer (150 mM NaCl, 50 mM Tris, 2.5 mM EDTA, 1% Triton X-100, 0.5% deoxycholic acid, 0.1% SDS, pH 8) with protease inhibitors, passed 10 times through a 33-gauge needle, and incubated for 1 h at room temperature with rotation. Insoluble material was pelleted by centrifugation at 20,000 ϫ g. Protein content in the supernatant was quantified using the BCA kit (Pierce), and protein (30 g/lane) was run on SDS-PAGE gels and transferred to nitrocellulose for Western blotting using standard techniques. To detect full-length APP and C-terminal fragments of APP, the blot was probed with CT15 (a generous gift from Dr. Ed Koo, University of California, San Diego, CA), a rabbit polyclonal antibody raised against the C-terminal 15 amino acids of APP (29). To detect LRP, blots were probed with rabbit anti-LRP antibody (17). Images were captured using the Kodak Im-ageStation 440CF, and densitometric analysis of bands was performed using the Kodak 1D Image Analysis software (Eastman Kodak Co., New Haven, CT).
Statistical Analysis-Data are reported as mean Ϯ S.E. and were analyzed using PRISM version 4.00 software (GraphPad, San Diego, CA).

RESULTS
To begin to determine whether one of the two major apoE receptors in the brain, LDLR or LRP, could regulate apoE levels in brain, we purified astrocyte-secreted, apoE-containing lipoprotein particles as described previously (18) and determined their ability to be endocytosed by mouse embryonic fibroblasts derived from wild-type, Ldlr Ϫ/Ϫ , Lrp Ϫ/Ϫ , or doubleknock-out Ldlr Ϫ/Ϫ , Lrp Ϫ/Ϫ mouse embryos (21). DiI-labeled apoE lipoprotein particles were readily taken up into endocytic vesicles in wild-type and Lrp Ϫ/Ϫ cells (Fig. 1A). This punctate pattern of staining is typical of receptor-mediated endocytosis and represents an endosomal/lysosomal distribution of endocytosed ligands. However, cells from Ldlr Ϫ/Ϫ or double-knock-out Ldlr Ϫ/Ϫ , Lrp Ϫ/Ϫ cells showed no appreciable endocytosis of astrocyte-secreted, apoE-containing lipoprotein particles (Fig.  1A). To further clarify the role of other LDLR family members in mediating the receptor-mediated endocytosis of astrocyte-derived apoE, we used Chinese hamster ovary cells that lack endogenous LRP and are stably transfected with the LDLR, LRP mini-receptors, apoER2, and the VLDL receptor to analyze potential uptake and degradation of apoE-containing lipoprotein particles (19). These cells expressing the different LDLR family members have been shown previously to bind and endocytose control ligands (19). These receptors are the main candidate apoE receptors expressed in the brain (15). Application of 125 I-labeled, astrocyte-secreted, apoE-containing lipoprotein particles to cells overexpressing LDLR resulted in a large increase in RAP-inhibitable endocytosis and degradation of apoE (Fig. 1B). However, cells overexpressing other LDLR family members or vector alone did not result in appreciable uptake and degradation of astrocyte-derived apoE (Fig. 1B).
To determine the potential role of LDLR in brain apoE metabolism in vivo, we used a sandwich enzyme-linked immunosorbent assay to measure the level of murine apoE in the extracellular space of the brain (such as the CSF) of both wild-type (Ldlr ϩ/ϩ ) and LDLR knock-out (Ldlr Ϫ/Ϫ ) mice. Because apoE is secreted into the brain extracellular space, we expected that if a receptor such as LDLR involved in its endocytosis was absent, extracellular levels would be elevated. For example, apoE is elevated severalfold in the plasma of Ldlr Ϫ/Ϫ mice (23) as well as in humans with familial hypercholesterolemia resulting from reduced LDLR expression and/or function (30,31). We found that Ldlr Ϫ/Ϫ mice had a 53% increase in CSF levels of apoE relative to Ldlr ϩ/ϩ mice ( Fig. 2A). Furthermore, we found that apoE levels in cortical tissue homogenized in PBS (as a reflection of extracellular pools of apoE) were significantly increased by 57% in Ldlr Ϫ/Ϫ mice compared with Ldlr ϩ/ϩ mice (Fig. 2B). Together, both the in vitro and in vivo data suggest that the LDLR plays an important role as an apoE receptor that mediates the uptake, degradation, and level of apoE in the brain.
Because Ldlr Ϫ/Ϫ mice have elevated extracellular levels of apoE in the brain and because apoE acts as an A␤ chaperone for both soluble and insoluble A␤, we wanted to determine the effect of LDLR on the deposition of A␤ in vivo. To determine whether LDLR has a direct effect on A␤ aggregation and deposition in vivo, we used the PDAPP mouse model of AD to generate PDAPP ϩ/Ϫ , Ldlr ϩ/ϩ and PDAPP ϩ/Ϫ , Ldlr Ϫ/Ϫ mice (littermates). PDAPP mice express the human APP transgene containing a familial AD mutation at amino acid 717 and overproduce A␤ (particularly A␤ 42 ). In the hippocampus of these mice, the amount of soluble and insoluble A␤ increases in an age-dependent manner beginning between 6 and 9 months of age, when plaques begin to deposit (25,32). We first examined the levels of both A␤ 40 and A␤ 42 in young PDAPP ϩ/Ϫ , Ldlr ϩ/ϩ and PDAPP ϩ/Ϫ , Ldlr Ϫ/Ϫ mice at 3 months of age, well before the deposition of A␤ begins. PBS-soluble levels of A␤ 40 and A␤ 42 in the hippocampus were not significantly different between the two genotypes (Fig. 3A). Carbonate-soluble levels of A␤ 40 and A␤ 42 in the hippocampus were also not significantly different between the two genotypes (Fig. 3B).
We next examined PDAPP ϩ/Ϫ , Ldlr ϩ/ϩ and PDAPP ϩ/Ϫ , Ldlr Ϫ/Ϫ mice at 10 months of age to determine the effect of the LDLR on A␤ deposition. The area of the hippocampus covered by A␤ immunoreactive deposits in tissue sections was 31% higher in 10-month-old PDAPP, Ldlr Ϫ/Ϫ mice compared with PDAPP, Ldlr ϩ/ϩ mice (Fig. 4A), but this increase was not statistically significant. Similar results were found for thioflavine-S-positive fibrillar deposits, and comparable results were found in the cortex (data not shown). A biochemical analysis of FIG. 1. LDLR regulates cellular uptake of astrocyte-secreted apoE. A, DiI-labeled, astrocyte-secreted apoE3-containing lipoprotein particles were readily endocytosed in wild-type and Lrp Ϫ/Ϫ mouse embryonic fibroblast cells, but not in Ldlr Ϫ/Ϫ or double-null Ldlr Ϫ/Ϫ , Lrp Ϫ/Ϫ mouse embryonic fibroblast cells. B, astrocyte-secreted apoE lipoprotein particles were labeled with 125 I and applied to Chinese hamster ovary LRP-null cells overexpressing LDLR family members. RAP-inhibitable binding of 125 I-labeled apoE3 was calculated and is presented as a percentage of pcDNA3.1 vector-only control. Only cells overexpressing the LDLR were able to endocytose and degrade apoE, as evidenced by trichloroacetic acid-soluble counts in the cell culture supernatant (percentage of control; ***, p Ͻ 0.001 by analysis of variance with post-hoc Tukey t test).

FIG. 2. LDLR regulates the level of apoE in the CNS.
A, the level of endogenous apoE in the CSF of 3-month-old mice was 1.84 Ϯ 0.08 g/ml in Ldlr Ϫ/Ϫ mice (n ϭ 8) versus 1.20 Ϯ 0.17 g/ml in Ldlr ϩ/ϩ mice (n ϭ 7) (**, p Ͻ 0.01 by unpaired two-tailed t test). B, the level of endogenous apoE in cortical PBS homogenates was 0.65 Ϯ 0.05 g/mg in Ldlr Ϫ/Ϫ mice (n ϭ 8) versus 0.41 Ϯ 0.08 g/mg in Ldlr ϩ/ϩ mice (n ϭ 7) (**, p Ͻ 0.01 by unpaired two-tailed t test). carbonate-soluble A␤ levels in the hippocampus revealed slight but nonsignificant increases in both A␤ 40 (17%) and A␤ 42 (53%) of 10-month-old PDAPP, Ldlr Ϫ/Ϫ mice as compared with PDAPP, Ldlr ϩ/ϩ mice (Fig. 4B). Carbonate-insoluble A␤ levels in the hippocampus also revealed slight but nonsignificant increases in both A␤ 40 (51%) and A␤ 42 (10%) of 10-month-old PDAPP, Ldlr Ϫ/Ϫ mice as compared with PDAPP, Ldlr ϩ/ϩ mice (Fig. 4C). Whereas Ldlr Ϫ/Ϫ mice have an increase in plasma cholesterol of ϳ79% compared with wild-type (Ldlr ϩ/ϩ ) mice (Fig. 5A), there were no differences in either brain or CSF cholesterol (Fig. 5, B and C). There was also no evidence that the processing of APP (Fig. 5D) or the levels of total APP protein (data not shown) were different between the genotypes. Additionally, hippocampal LRP levels were not altered in the absence of LDLR (data not shown). Thus, although the level of murine apoE is elevated by about 50% in the extracellular CNS pools in Ldlr Ϫ/Ϫ mice, these data suggest that this degree of change in the level of apoE is not sufficiently increased to significantly affect A␤ levels at young ages or the early stages of A␤ deposition in PDAPP transgenic mice.
To explore the possibility that the LDLR may differentially regulate the level of human apoE isoforms in the CNS compared with murine apoE, we bred human APOE2, APOE3, and APOE4 targeted replacement mice (knock-in) onto an Ldlr Ϫ/Ϫ mouse background. To confirm previous findings noted from these mice (33), we assessed plasma cholesterol and apoE and found that both were significantly elevated in the absence of LDLR in all genotypes (data not shown). To determine the extent to which LDLR regulates human apoE levels in the extracellular space of the CNS, we measured apoE levels in the CSF. We found that levels of both human apoE3 and human apoE4 in the CSF were significantly higher by 210% and 380%, respectively, in the absence of the LDLR (Fig. 6A). The level of human apoE2 in the CSF was not significantly altered by the presence or absence of the LDLR (Fig. 6A), as was expected, because the human apoE2 isoform exhibits Ͻ2% FIG. 4. LDLR does not affect A␤ deposition in PDAPP mice at 10 months of age. A, the percentage of cross-sectional area of the hippocampus covered with A␤ immunoreactive deposits was not significantly different in PDAPP, Ldlr ϩ/ϩ mice (1.66%) compared with PDAPP, Ldlr Ϫ/Ϫ mice (2.18%). B, the level of carbonate-soluble A␤ 40 was not significantly different in PDAPP, Ldlr ϩ/ϩ mice (0.312 Ϯ 0.085 ng/mg) compared with PDAPP, Ldlr Ϫ/Ϫ mice (0.366 Ϯ 0.072 ng/mg). The level of carbonate-soluble A␤ 42 was also not significantly different in PDAPP, Ldlr ϩ/ϩ mice (4.17 Ϯ 0.66 ng/mg) compared with PDAPP, Ldlr Ϫ/Ϫ mice (6.40 Ϯ 1.00 ng/mg). C, the level of carbonate-insoluble A␤ 40 was not significantly different in PDAPP, Ldlr ϩ/ϩ mice (6.24 Ϯ 0.80 ng/mg) compared with PDAPP, Ldlr Ϫ/Ϫ mice (9.45 Ϯ 1.22 ng/mg). The level of carbonate-soluble A␤ 42 was also not significantly different in PDAPP, Ldlr ϩ/ϩ mice (262 Ϯ 58 ng/mg) compared with PDAPP, Ldlr Ϫ/Ϫ mice (288 Ϯ 47 ng/mg).
binding to the LDLR (34). However, it is interesting to note that the level of human apoE2 is substantially higher than that of apoE3 or apoE4 in the CSF (Fig. 6A). Despite the increase in human apoE levels in the absence of LDLR, there was no significant increase in the level of brain cholesterol in these mice (Fig. 6B). DISCUSSION In the present study, we found that the LDLR, but not other LDLR family members, was able to efficiently bind, endocytose, and degrade astrocyte-secreted apoE-containing lipoprotein particles and that Ldlr Ϫ/Ϫ mice have ϳ50% higher levels of apoE in CSF and extracellular pools of brain tissue. However, when the PDAPP mouse model of AD was bred onto an Ldlr Ϫ/Ϫ background, there was not a significant increase in the levels of A␤ 40 and A␤ 42 at young ages before deposition of A␤ begins. Additionally, there was not a significant increase in A␤ deposition defined by immunohistochemical and biochemical measures between 10-month-old PDAPP, Ldlr Ϫ/Ϫ and PDAPP, Ldlr ϩ/ϩ mice. Interestingly, using human APOE knock-in mice, we found that the increase in the level of apoE3 and apoE4 (but not apoE2) on an Ldlr Ϫ/Ϫ background was much greater than that for murine apoE on this same Ldlr Ϫ/Ϫ background.
Our in vitro data indicate that LDLR family members including LRP, VLDL receptor, and apoER2 are not able to appreciably internalize and degrade astrocyte-secreted apoE-containing lipoproteins. Previous work has clearly demonstrated that these LDLR family members are able to efficiently bind and endocytose apoE reconstituted in large, cholesterol-rich lipoprotein particles termed ␤-VLDL (35). However, these ␤-VLDL particles differ from astrocyte-secreted apoE in lipid composition, the presence of other apoproteins, and the amount of sialation (12). The apolipoprotein lipidation state does alter receptor binding characteristics. For example, our previous studies have shown that recombinant apoE in the absence of lipid prefers binding to LRP over the LDLR (22). This is not a form of apoE that has been shown to be present under physiological conditions. Together, these results indicate that the lipid content and form of apoE lipoprotein particles can alter their receptor binding specificity. They also demonstrate that the LDLR is an important apoE receptor that regulates human and murine apoE endocytosis and levels in the brain. Although FIG. 6. LDLR influences the levels of human apoE3 and apoE4 but not apoE2 in the cerebrospinal fluid. A, the level of apoE in the cerebrospinal fluid was measured in human APOE2 ϩ/ϩ , APOE3 itsup;ϩ/ϩ, and APOE4 ϩ/ϩ targeted replacement mice (knock-in) in the presence (Ldlr ϩ/ϩ ) or absence (Ldlr Ϫ/Ϫ ) of the endogenous LDLR. The level of human apoE was significantly higher in APOE2 ϩ/ϩ mice (5.54 Ϯ 0.78 g/ml) compared with APOE3 ϩ/ϩ (1.94 Ϯ 0.25 g/ml) and APOE4 ϩ/ϩ (1.28 Ϯ 0.26 g/ml) mice (a, p Ͻ 0.001 by analysis of variance with post-hoc Tukey t test). As expected from previous studies, the level of human apoE was not significantly different between APOE2 ϩ/ϩ , Ldlr ϩ/ϩ and APOE2 ϩ/ϩ , Ldlr Ϫ/Ϫ mice. The level of human apoE was significantly higher in APOE3 ϩ/ϩ , Ldlr Ϫ/Ϫ mice (4.04 Ϯ 0.51 g/ml) compared with APOE3 ϩ/ϩ , Ldlr ϩ/ϩ mice (1.94 Ϯ 0.25 g/ml) (*, p Ͻ 0.05 by analysis of variance with post-hoc Tukey t test). The level of human apoE was also significantly higher in APOE4 ϩ/ϩ , Ldlr Ϫ/Ϫ mice (4.81 Ϯ 0.47 g/ml) compared with APOE3 ϩ/ϩ , Ldlr ϩ/ϩ mice (1.28 Ϯ 0.26 g/ml) (***, p Ͻ 0.001 by analysis of variance with post-hoc Tukey t test). B, the level of total cholesterol in the cortex was not increased in human APOE2 ϩ/ϩ , APOE3 ϩ/ϩ , and APOE4 ϩ/ϩ knock-in mice in the absence of the LDLR. Interestingly, human APOE3 ϩ/ϩ had a slight but significant decrease in the level of cholesterol in the cortex in the absence of the LDLR (*, p Ͻ 0.05).
FIG. 5. LDLR influences levels of plasma cholesterol but has no effect on hippocampal and CSF cholesterol or APP processing. A, the level of total plasma cholesterol was 69.2 Ϯ 4.63 mg/dl in Ldlr ϩ/ϩ mice (n ϭ 6) as compared with 124 Ϯ 8.82 mg/dl in Ldlr Ϫ/Ϫ mice (n ϭ 6) (***, p Ͻ 0.001, unpaired two-tailed t test). Level of total cholesterol in the hippocampus (B), CSF (C), and APP processing as assessed by semiquantitative Western blotting of ␣-C-terminal fragment (␣-CTF; D) was not significantly different between the genotypes (n ϭ 5-7/group). the results of this study do not rule out that other LDLR family members contribute in some way to the level of apoE in the brain, they clearly show that LDLR plays an important role in brain apoE metabolism.
Recent studies have shown that increasing or decreasing cellular cholesterol can influence cellular A␤ secretion (36). Although Ldlr Ϫ/Ϫ mice have elevated plasma cholesterol levels, we found that the level of brain cholesterol is unaffected in these mice as has been previously reported (37). The exact mechanism by which apoE affects AD pathogenesis is unclear. However, a large body of evidence suggests that one mechanism by which it influences the age of onset of AD is by acting as a chaperone for both soluble and insoluble A␤, thereby influencing A␤ clearance and fibrillogenesis (3,5,7,32). Our current results suggest that although the LDLR regulates extracellular levels of apoE in the brain, a 50% increase in the level of murine apoE over endogenous levels is not sufficient to alter A␤ levels at young ages or the early stages of A␤ deposition in PDAPP mice at 10 months of age. This may represent a ceiling effect of the level of murine apoE or indicate that a larger increase in the level of murine apoE may be necessary to effect a change in A␤ levels in this model. Alternatively, other pathways of apoE-mediated clearance of A␤ play a much larger role in the metabolism of A␤ such as LRP or bulk flow along interstitial fluid drainage pathways. Unfortunately, Lrp Ϫ/Ϫ mice die in early embryonic stages, and thus the effect of endogenous LRP on brain apoE levels is difficult to assess as shown in this study. Additionally, the LDLR may play a role in A␤ deposition only at later stages of deposition, as was found for LRP in PDAPP mice (17). It is also possible that the LDLR may differentially regulate human apoE isoforms in the CNS to a greater extent than murine apoE and therefore influence A␤ levels and deposition. It is interesting that independent of the effect of the LDLR, the level of apoE in the CSF of human APOE2 mice is significantly higher than that in APOE3 or APOE4 mice. This suggests that in humans, LDLR may play a more important role in regulating CNS apoE levels and AD pathogenesis. Studies of human CSF have reported no significant differences in the level of human apoE between individuals expressing APOE3 or APOE4 alleles, but the level of CSF apoE in APOE2/2 homozygous individuals has not been assessed, to our knowledge. Given that APOE2 has been associated with a decreased risk for developing AD, an intriguing possibility is that the absolute level of apoE present in the brain of these individuals could account for the decrease in risk. Human and murine apoE differ by ϳ25% at the amino acid level, and human apoE isoforms and murine apoE isolated from VLDL have been shown to differentially bind the LDLR (33). Using the PDAPP mouse model, previous studies from our group have shown that expression of human apoE isoforms under the control of the glial fibrillary acidic protein promoter results in a significant delay in deposition compared with mice expressing endogenous murine apoE (7,32). More recently, the same human apoE knock-in mice used in this study were crossed to PDAPP mice, and both a delay in A␤ deposition and an isoform-dependent effect on A␤ deposition (E4 Ͼ E3 Ն E2) were found (38). Additionally, we have recently replicated this result in the APPsw mouse model of AD utilizing the same APOE3 and APOE4 knock-in mice used in this study (39). Clearly, human apoE isoforms affect A␤ metabolism and deposition differently than murine apoE, but whether this is due to altered clearance and/or structure of A␤ in vivo remains to be determined. Determining the role of LDLR or other receptors or clearance pathways in the brain that are also involved in regulating apoE levels is central to understanding how apoE influences A␤ metabolism and AD pathogenesis.