Interaction of Apolipoprotein J-Amyloid β-Peptide Complex with Low Density Lipoprotein Receptor-related Protein-2/Megalin

Apolipoprotein J (apoJ) has been shown to be the predominant amyloid β-peptide (Aβ)-binding protein in cerebrospinal fluid. We have previously demonstrated that the endocytic receptor low density lipoprotein receptor-related protein-2/megalin (LRP-2), which is expressed by choroid plexus epithelium and ependymal cells lining the brain ventricles and neural tube, binds and mediates cellular uptake of apoJ (Kounnas, M. Z., Loukinova, E. B., Stefansson, S., Harmony, J. A., Brewer, B., Strickland, D. K., and Argraves, W. S. (1995) J. Biol. Chem. 270, 13070–13075). In the present study, we evaluated the ability of apoJ to mediate binding of Aβ1–40-apoJ complex to LRP-2 in vitro. Immunoblot analysis showed that incubation of apoJ with Aβ1–40 resulted in the formation of Aβ1–40-apoJ complex and the inhibition of the formation of Aβ1–40 aggregates. Using an enzyme-linked immunosorbent assay, an estimated dissociation constant (K d ) of 4.8 nm was derived for the interaction between Aβ1–40 and apoJ. Enzyme-linked immunosorbent assay was also used to study the interaction of the Aβ1–40-apoJ complex with LRP-2. The results showed that Aβ alone did not bind directly to LRP-2; however, when Aβ1–40 was combined with apoJ to form a complex, binding to LRP-2 took place. The binding interaction could be blocked by inclusion of the receptor-associated protein, an antagonist of apoJ binding to LRP-2. When LRP-2-expressing cells were given125I-Aβ1–40, cellular uptake of the radiolabeled peptide was promoted by co-incubation with apoJ. When the cells were provided purified125I-Aβ1–40-apoJ complex, the complex was internalized and degraded, and both processes were inhibited with polyclonal LRP-2 antibodies. Furthermore, chloroquine treatment inhibited the cellular degradation of the complex. The data indicate that apoJ facilitates Aβ1–40 binding to LRP-2 and that the receptor mediates cellular clearance of Aβ1–40-apoJ complex leading to lysosomal degradation of Aβ1–40. The findings support the possibility that LRP-2 can act in vivoto mediate clearance of the complex from biological fluids such as cerebrospinal fluid and thereby play a role in the regulation of Aβ accumulation.

A hallmark feature of Alzheimer's disease is the accelerated cerebral accumulation of amyloid ␤-protein (A␤), 1 a small 39 -42-residue proteolytically derived fragment of amyloid ␤-precursor protein (1,2). The accumulation takes the form of spherical extracellular deposits of A␤ fibrils in the vicinity of morphologically abnormal axons and dendrites. Associated with these so-called plaques are microglia and astrocytes. The mechanisms that lead to accumulation of A␤ are still obscure but represent an area of intense investigation. Whereas much emphasis is currently placed on trying to determine the mechanism(s) of A␤ biosynthesis from amyloid ␤-precursor protein processing (3,4), little is being done on determining possible mechanisms that mediate the catabolism of A␤. Catabolic processes may prevent the extracellular accumulation of A␤ that is expressed under normal physiological conditions yet does not accumulate to the extent seen in Alzheimer's disease or Down's syndrome.
A␤ can be found in cerebrospinal fluid and blood in complex with apolipoprotein J (apoJ) or apolipoprotein E (apoE) (5,6). Whereas apoE has been reported to promote A␤ fibrilogenesis (7)(8)(9), apoJ has been shown to slow the formation of A␤ aggregates and may therefore act to maintain A␤ in a soluble form and prevent it from forming pathological fibrils (10). Our discoveries that LRP-2 2 is an endocytic receptor for apoJ (11) and LRP-2 is expressed by cells that are in contact with cerebrospinal fluid (choroid plexus and ependymal cells) (12) prompted us to hypothesize that LRP-2 may mediate clearance of A␤ complexed with apoJ, thereby controlling the accumulation of A␤. In the present study, we used in vitro solid phase binding assays as well as cellular internalization and degradation assays to evaluate the roles of apoJ and LRP-2 in mediating cellular clearance of A␤.

EXPERIMENTAL PROCEDURES
Proteins-Human apoJ was purchased from Quidel (San Diego, CA). Synthetic A␤ fragment 1-40 and ovalbumin were obtained from Sigma. Bovine serum albumin was purchased from U. S. Biochemical Corp. Human RAP was expressed as a glutathione S-transferase fusion protein in bacteria and prepared free of glutathione S-transferase as described by Williams et al. (13). LRP-2 was purified from extracts of porcine kidney by affinity chromatography using a column of RAP coupled to Sepharose as described previously (36).
Antibodies-The mouse monoclonal antibody to LRP-2 designated 1H2 was provided by Dr. Robert McCluskey (Massachusetts General Hospital, Boston, MA). Mouse monoclonal antibody to human apoJ (mAb 1D11) was obtained from Dr. Judith Harmony (University of Cincinnati College of Medicine, Cincinnati, OH). Mouse monoclonal antibody to human A␤ (mAb 4G8) was purchased from Senetek (Maryland Heights, MO). Rabbit anti-LRP-2 IgGs (rabbit 6286) were isolated by immunoaffinity chromatography on a column of porcine LRP-2 coupled to CNBr-activated Sepharose (Pharmacia Biotech Inc.) with minor modification to a previously described procedure (11). IgG was sequentially eluted using 100 mM glycine, pH 2.3, followed by 100 mM triethylamine, pH 11.5, and the combined eluates were dialyzed against 50 mM Tris, pH 7.4, 150 mM NaCl. The polyclonal anti-LRP-2 IgG preparation was absorbed on a column of RAP-Sepharose followed by selection on a column of protein G-Sepharose. Control rabbit IgG was isolated from the preimmune serum of rabbit 6286 by protein G-Sepharose chromatography.
Formation of A␤-ApoJ Complex-ApoJ was combined with synthetic A␤ 1-40 at a 1:15 molar ratio in PBS as described previously (14) and incubated for 24 h at 37°C. Typically, a 1:15 molar ratio was maintained for preparing complexes of unlabeled A␤ 1-40 and apoJ, although the total protein concentration varied according to the assay. For SDS-PAGE and immunoblotting analyses the concentration of apoJ was 0.95 M and A␤ 1-40 was 15 M, whereas for solid phase binding assays, apoJ was 0.095 M and A␤ 1-40 was 1.5 M. As a control, ovalbumin was substituted for apoJ.
Enzyme-linked Immunosorbent Assays (ELISA)-Microtiter wells were coated with LRP-2, A␤, or BSA (each at 3 g/ml) in 150 mM NaCl, 50 mM Tris, pH 8.0 (TBS) containing 5 mM CaCl 2 for 18 h at 4°C. Unoccupied sites were either blocked with TBS containing 3% nonfat milk or treated with PBS containing 0.1% N-octyl-␤-D-glucopyranoside (OG) (Calbiochem). For those wells that were blocked with TBS containing 3% nonfat milk, all subsequent incubations were carried out in TBS containing 3% nonfat milk plus 0.1% OG. For wells treated with PBS containing 0.1% OG, all subsequent incubations were carried out in PBS containing 0.1% OG. Bound proteins were detected using mouse monoclonal antibodies, sheep anti-mouse IgG-horseradish peroxidase (Amersham International, Buckinghamshire, UK), and the chromogenic substrate o-phenylenediamine (Sigma) (1 mg/ml in 12 mM citric acid monohydrate, 25 mM dibasic sodium phosphate, pH 5.0, 0.0014% hydrogen peroxide). Binding data from ELISA were analyzed using a form of the binding isotherm as described by Ashcom et al. (15).
Immunoblotting-To analyze A␤-apoJ complexes by SDS-PAGE, samples were electrophoresed on 10 -20% acrylamide gradient, Tricinecontaining gels (Novex) in the presence of SDS. The separated proteins were electrophoretically transferred to Protran TM nitrocellulose membranes (Schleicher & Schuell) in Tris/glycine-methanol buffer for 2 h at 70 V. After transfer, the membranes were incubated with 5% nonfat dry milk in TBS (pH 7.4). The membranes were then incubated with monoclonal A␤ or apoJ antibodies followed by sheep anti-mouse IgG-horseradish peroxidase (Amersham) diluted in 5% nonfat milk, TBS, 0.1% Tween 20. To detect bound antibodies, the membranes were incubated with ECL TM Western blotting chemiluminescent reagent (Amersham) and exposed to Biomax TM MR film (Eastman Kodak Co.). 125 I-A␤-ApoJ Complex-To evaluate the effect of apoJ on the internalization of 125 I-A␤, mouse teratocarcinoma F9 cells (ATCC CRL 1720) were treated for 6 days with retinoic acid (RA) and dibutyryl cyclic AMP (Bt 2 cAMP) as described previously (30), released by trypsin-EDTA, and reseeded onto gelatin-coated 383-mm 2 wells of 12-well plates (1.5 ϫ 10 5 cells/well) in Dulbecco's modified Eagle's medium (DMEM, Life Technologies, Inc.) containing 10% bovine calf serum (Hyclone Laborato-ries), 20 mM HEPES, pH 7.4, 100 units/ml penicillin, 100 g/ml streptomycin (Life Technologies, Inc.) and no RA/Bt 2 cAMP. The cells were cultured for 18 h at 37°C, 5% CO 2 and then washed with serum-free DMEM containing penicillin and streptomycin. 125 I-A␤ (70 nM) in DMEM, 1.5% BSA, 1% Nutridoma serum substitute (Boehringer Mannheim) (DMEM/BSA/SS) containing various amounts of apoJ (1-40 nM) was added to the cells and incubated for 5 h at 37°C, 5% CO 2 . The amount of radiolabeled A␤ that was internalized by cells was defined as the amount of radioactivity that remained associated with the cell pellets following trypsin/proteinase K/EDTA treatment (11,30).

Assay of Cellular Internalization and Degradation of 125 I-A␤ and
To evaluate the role of LRP-2 in the cellular internalization and degradation of 125 I-A␤-apoJ complex, RA/Bt 2 cAMP-treated mouse teratocarcinoma F9 cells were cultured as described above. 60 min prior to the addition of 125 I-A␤-apoJ complex, the medium was removed and the cells were treated with DMEM/BSA/SS containing either anti-LRP-2 IgG (200 g/ml), control rabbit IgG (200 g/ml), or chloroquine (0.1 mM). 125 I-A␤-apoJ complex (10 nM) in DMEM/BSA/SS or in DMEM/BSA/SS containing anti-LRP-2 IgG (200 g/ml), rabbit IgG (200 g/ml), or chloroquine (0.1 mM) was then added and incubated with the cells for 5 or 18 h at 37°C, 5% CO 2 . The amount of radiolabeled complex that was internalized was measured as described above. Radioactivity released into the conditioned culture medium that was soluble in 10% trichloroacetic acid was taken to represent degraded ligand. Total ligand degradation values were corrected for non-cellular mediated degradation by subtracting the amount of degradation that occurred when the radiolabeled complex was incubated in wells lacking cells.

RESULTS
Amyloid ␤-Protein Binds to ApoJ but Not to LRP-2-ELISAs were used to determine whether synthetic A␤ 1-40 peptide was capable of binding directly to purified LRP-2. As shown in Fig.  1A, LRP-2 did not bind to microtiter wells coated with A␤ 1-40 . Likewise, A␤ 1-40 (at concentrations up to 125 nM) did not bind to LRP-2-coated wells (Fig. 1B). In parallel assays, A␤ 1-40 was shown to bind to apoJ in a dose-dependent manner, either when A␤ 1-40 was coated onto microtiter wells and apoJ was introduced in solution phase (Fig. 1A) or when apoJ was coated and A␤  was introduced in solution phase (Fig. 1B). The data for solution-phase apoJ binding to immobilized A␤ 1-40 (Fig. 1A) were fit using a hyperbolic function (16), and the half-saturating level of binding (estimated K d ) was determined to be 4.8 nM. This value is in good agreement with the value of 2.0 nM reported by Matsubara et al. (6). By contrast, the binding of solution-phase A␤ 1-40 to immobilized apoJ was not saturable (Fig. 1B). Such non-saturable binding can be expected given that A␤ 1-40 has the ability to self-associate (17). The results indicate that A␤ 1-40 does not bind to LRP-2 but does bind with high affinity to apoJ.
Amyloid ␤-Protein-ApoJ Complex Binds to LRP-2-To generate A␤-apoJ complex, A␤ 1-40 was incubated with apoJ for 24 h at 37°C, and complex formation was evaluated by SDS-PAGE and immunoblot analysis. As shown in Fig. 2, this incubation resulted in the formation of an approximately 70-kDa band, immunoreactive with monoclonal A␤ antibody (Fig. 2B,  lane 4). The anti-A␤-reactive 70-kDa band displayed a similar electrophoretic mobility to the 70-kDa apoJ (Fig. 1C, compare  lanes 4 and 6). Incubation of A␤ with ovalbumin did not produce a band having a similar molecular mass (Fig. 2B, lane 5).
Coomassie Blue staining showed that in the lane containing A␤, which had been incubated alone for 24 h at 37°C, there was a single band having a mobility corresponding to ϳ4 kDa ( Fig.  2A, lane 3), consistent with M r of 4392. However, immunoblot analysis of this lane (Fig. 2B, lane 3) using monoclonal A␤ antibody revealed several immunoreactive species having M r values of ϳ8000 and ϳ12,000 and a high molecular mass band that just entered the gel. These species likely correspond to A␤ dimer, trimer, and aggregate, respectively. Although the dimer, trimer, and aggregated species were not detectable by Coomassie Blue staining, they were immunoreactive with A␤ antibody, indicative perhaps of its preference for multimerized peptide versus the monomeric form. Each of these immunoreactive species was also present in the profile of the A␤ incubated with ovalbumin. However, the A␤ aggregate was missing in the lane containing A␤ incubated with apoJ (Fig. 2B, lane 4). The data indicated that incubation of apoJ with A␤ under the conditions that we described resulted in the formation of a complex of A␤ and apoJ that is stable in SDS. The conditions of SDS-PAGE were insufficient to permit resolution of the A␤-apoJ complex as a discrete species having a M r ϳ4000 greater than apoJ. The results also showed that apoJ inhibited the formation of aggregated A␤ while not perturbing the formation of A␤ dimer and trimer.
To evaluate the ability of A␤-apoJ complex to interact with LRP-2, microtiter wells coated with LRP-2 were incubated with mixtures of A␤ and apoJ or A␤ and ovalbumin that had been preincubated for 24 h at 37°C. As shown in Fig. 3A, A␤ binding to immobilized LRP-2, as detected by A␤ monoclonal antibody, occurred when A␤ was preincubated with apoJ but not with ovalbumin. Experimental controls showed that neither of the mixtures, A␤ and apoJ or A␤ and ovalbumin, bound to wells coated with BSA (Fig. 3B). When monoclonal antibody to apoJ was used to detect apoJ binding to LRP-2, the half-saturating level of binding of apoJ to LRP-2 was not significantly modified by the inclusion of A␤ (Fig. 3C). The results indicate that apoJ mediates binding of A␤ to LRP-2 and that the affinity of the A␤-apoJ complex for LRP-2 does not appear to be different from that of apoJ alone.
The Antagonist of ApoJ Binding to LRP-2 Blocks Binding of A␤-ApoJ Complex to LRP-2-RAP has been shown to inhibit the binding of apoJ to LRP-2 (11). As shown in Fig. 4, incubation of A␤-apoJ complex with RAP completely blocked the binding of the complex to immobilized LRP-2. This, taken together with the above data, indicates that apoJ can function to bridge the interaction of A␤ with LRP-2.
Cellular Endocytosis of A␤ Is Facilitated by ApoJ-To determine whether apoJ might facilitate cellular internalization of A␤, 125 I-A␤ was administered to cultured LRP-2-expressing cells in the presence of varying concentrations of apoJ. As shown in Fig. 5, exogenously added apoJ promoted the internalization of 125 I-A␤ in a dose-dependent manner.
LRP-2 Mediates Cellular Endocytosis and Degradation of A␤-ApoJ Complex-We next examined the cellular clearance of A␤-apoJ complex and evaluated the role of LRP-2 in the process. Radioiodinated A␤ was combined with unlabeled apoJ, and the resulting complex was purified by gel filtration chromatography. Fig. 6A shows the chromatographic profiles of the individual components and the complex-containing mixture. Native gel electrophoretic analysis of the complex-containing fraction is shown in the inset of Fig. 6A. The results indicated that the chromatography procedure permitted isolation of the 125 I-labeled complex from the bulk of the unincorporated A␤; however, free radiolabeled peptide did copurify with the complex.
Equimolar amounts of 125 I-A␤ or 125 I-A␤-apoJ were administered to LRP-2-expressing cells, and the level of internalization of each was measured. As shown in Fig. 6B, there was a 2-fold higher level of 125 I-A␤-apoJ complex internalized as compared with 125 I-A␤ alone. In separate experiments, the amount of 125 I-A␤ internalized was unchanged by the inclusion of a 1000-fold molar excess of unlabeled peptide (data not shown). Complex internalization was also measured in the presence of function blocking LRP-2 antibodies or control IgGs. As shown in Fig. 6B (and Fig. 7A), anti-LRP-2 IgG inhibited 125 I-A␤-apoJ complex internalization by 59% as compared with treatment with control rabbit IgGs. The results indicate that A␤-apoJ complex is internalized by LRP-2-expressing cells to a greater extent than A␤ alone and that the internalization of the complex can be inhibited by LRP-2 antibodies, implicating LRP-2 in the clearance process.
We also evaluated whether the internalized complex was lysosomally degraded as is the case for other LRP-2 ligands including apoJ (11). As shown in Fig. 7, administration of 125 I-A␤-apoJ complex to LRP-2-expressing cells resulted in the internalization and degradation of the complex. Degradation was evidenced by the appearance of trichloroacetic acid-soluble radioactivity in the conditioned culture medium that could be blocked by treatment with chloroquine, an inhibitor of lysoso- mal proteinase activity (Fig. 7B). Both internalization and degradation of the complex were also inhibited with LRP-2-antibodies. Taken together, the findings indicate that LRP-2 mediates endocytosis of A␤-apoJ complex leading to its degradation in lysosomes.

DISCUSSION
In this study, we document the ability of the A␤-apoJ complex to bind to the endocytic receptor LRP-2 in both cell-free and cultured cell assays. Although the physiological relevance of this interaction remains to be established, we have previously hypothesized that it is part of a mechanism in which LRP-2-expressing epithelial cells, such as those of the choroid plexus and ependyma, can clear the A␤-apoJ complex from the cerebrospinal fluid (11). In support of such a hypothesis is the fact that both apoJ and LRP-2 are expressed at high levels in the choroid plexus epithelium as well as ependymal cells that line the ventricles of the brain and neural tube (12,18,19) and are therefore in direct contact with cerebrospinal fluid. In addition to LRP-2 having a possible role in clearance of A␤-apoJ complex from the cerebrospinal fluid, there is evidence that LRP-2 may have a role in uptake of the complex from the blood by cells of the cerebral vascular endothelium (14,20). Following our previous report on the identification of LRP-2 as the receptor for apoJ, Zlokovic et al. (14) introduced radiolabeled A␤-apoJ complex into rat brain vasculature and observed that RAP or monoclonal antibody to LRP-2 decreased the brain uptake of the radiolabeled complex. Whereas these findings indirectly implicate LRP-2 as being responsible for the observed vascular clearance, evidence is needed to show that LRP-2 is indeed expressed by brain vascular endothelial cells. Nevertheless, the results presented in the present study provide direct evidence that LRP-2 is a receptor for the A␤-apoJ complex, thus establishing that it could serve to mediate A␤-apoJ endocytosis in any LRP-2-expressing tissue.
Our results also indicate that apoJ inhibits the formation of high molecular weight aggregates of A␤   (Fig. 2). This activity is consistent with observations made using a sedimentation assay (10) and with the hypothesis that apoJ serves to maintain A␤ in a soluble form, preventing it from forming insoluble amyloid filaments. Such filaments are the hallmark component of the senile plaque found in brain parenchyma and deposited in the cerebrovasculature of patients with Alzhei- mer's disease (21,22). In addition, the A␤-apoJ interaction may have a cytoprotective effect given that the insoluble aggregated form of A␤ has been shown to be toxic to cultured neuronal cells (23). By contrast to the anti-amyloidogenic and cytoprotective actions of apoJ, apoE has been shown to promote the formation of A␤ fibrils (7,9). Further studies are required to understand the factors that affect the in vivo interaction of A␤ with apoJ or apoE. It is possible that under normal physiological conditions, the interaction between apoJ and A␤ is favored over that of apoE to prevent amyloidogenesis. Moreover, clearance of A␤-apoE by endocytic apoE receptors such as LRP-1 and LRP-2 may also function to limit the extracellular level of complex as we have proposed for the A␤-apoJ complex. In this regard, the endocytic action of LRP-1 and the recently described low density lipoprotein receptor family member termed apoER2 (24) may be more relevant to clearance of A␤-apoE complex from brain parenchyma than LRP-2 since LRP-1 and apoER2 have widespread expression in the parenchyma (24 -27), whereas LRP-2 expression in the brain is restricted to epithelial cells of the choroid plexus and ependyma (12,28,29).
Herein, evidence is presented indicating that one conse-quence of LRP-2-mediated endocytosis of A␤-apoJ is that A␤ is targeted for lysosomal degradation. This is the end result for other LRP-2 ligands following their endocytosis (e.g. urokinase and plasminogen activator inhibitor-1 complex, low density lipoprotein, and apoJ (11, 30 -32)). It may, however, seem paradoxical that A␤ could be degraded in lysosomes, considering that a number of studies indicate that it can be created in lysosomes as a result of proteolytic processing of amyloid ␤-precursor protein (33,34). It is possible that the lysosomal presentation of A␤ in the form of a complex with apoJ may make it more susceptible to proteinase degradation. It is important to note that LRP-2 is apparently not the only means by which extracellular A␤ can be internalized. For example, fibroblasts presumably lacking LRP-2 can internalize exogenously added 125 I-A␤ 1-42 (35). This internalization pathway was shown to lead to intracellular accumulation of A␤ in the form of aggregates that are resistant to lysosomal proteinase degradation. LRP-2-mediated internalization may avoid this outcome by both preventing intracellular aggregate formation and promoting lysosomal degradation of A␤.