Expression of Scavenger Receptors in Glial Cells

Astrocytes and microglia associate to amyloid plaques, a pathological hallmark of Alzheimer disease. Microglia are activated by and can phagocytose β-amyloid (Aβ). Scavenger receptors (SRs) are among the receptors mediating the uptake of fibrillar Aβ in vitro. However, little is known about the function of the astrocytes surrounding the plaques or the nature of their interaction with Aβ. It is unknown whether glial cells bind to nonfibrillar Aβ and if binding of astrocytes to Aβ depends on the same Scavenger receptors described for microglia. We determined the binding of glia to Aβ by an adhesion assay and evaluated the presence of scavenger receptors in glial cells by immunocytochemistry, immunohistochemistry of brain sections, and immunoblot. We found that astrocytes and microglia from neonatal rats adhered in a concentration-dependent manner to surfaces coated with fibrillar Aβ or nonfibrillar Aβ. Fucoidan and poly(I), known ligands for SR-type A, inhibited adhesion of microglia and astrocytes to Aβ and also inhibited Aβ phagocytosis. In contrast, a ligand for SR-type B like low density lipoprotein, did not compete glial adhesion to Aβ. Microglia presented immunodetectable SR-BI, SR-AI/AII, RAGE, and SR-MARCO (macrophage receptor with collagenous structure, a member of the SR-A family). Astrocytes presented SR-BI and SR-MARCO. To our knowledge, this is the first description of the presence of SR-MARCO in astrocytes. Our results indicate that both microglia and astrocytes adhere to fibrillar and nonfibrillar Aβ. Adhesion was mediated by a fucoidan-sensitive receptor. We propose that SR-MARCO could be the Scavenger receptor responsible for the adhesion of astrocytes and microglia to Aβ.

Scavenger receptors have been implicated in microglial binding to A␤. However, their participation has not been adequately characterized in astrocytes. Astrocytes serve important metabolic and supportive functions in the central nervous system. Under physiological conditions, they maintain the functional integrity of synapses and produce some of the glycoproteins of the extracellular matrix (9). They regulate glutamate and water balance and have a role in the maintenance of the bloodbrain barrier tight junctions. Astrocytes express several chemokines, cytokines, and their receptors and have phagocytic and proteolytic activity (10 -12). Whether normal adult astrocytes can perform phagocytosis remains an unresolved issue. However, there is some evidence that under certain conditions astrocytes are capable of phagocytosis (13,14). In contrast to microglia, only SR-BI seems to be expressed by astrocytes from adult mouse and human brains (15), although there are reports suggesting that astrocytes also express the mannose receptor (16). Under pathological conditions, like injury of the nervous system, the activation of astrocytes has been associated with changes in the expression of cytoplasmatic proteins, surface proteins, and soluble factors, such as chemokines (10,12).
Alzheimer disease (AD) is a degenerative disorder characterized by selective neuronal loss and the presence of two types of protein aggregates, amyloid plaques, and neurofibrillary tangles (17,18,19,20). Amyloid plaques are complex deposits composed mainly of fA␤ (21,22) closely associated with microglia and astrocytes (23)(24)(25). fA␤ is neurotoxic in vitro (26). In AD brains, microglia are activated at the site of deposition of A␤ (27), and astrocytes are most closely associated with plaques with a dense amyloid core and activated microglia (28,29). Activated astrocytes surrounding amyloid plaques or exposed to fA␤ in vitro increase the expression of inducible nitricoxide synthase (30) and also secrete proinflammatory molecules such as interleukin-1␤, MCP-1, RANTES, etc. (10,31). It is unknown which receptor mediates the activation of astrocytes by A␤.
SR-MARCO is a newly described SR. It was initially characterized as a bacteria binding receptor. However, it also binds modified LDL; therefore, it is considered a member of the SR-A family (32,33). There is little information regarding the expression of SR-MARCO in the central nervous system, although its presence in microglial cells was recently described (34,35).
Recent studies showed that astrocytes from adult mice bind A␤ by receptors sharing characteristics with SRs. However, the results suggested that SR-BI did not bind fA␤ (36). Furthermore, the binding of microglial cells and astrocytes to fibrillar versus nonfibrillar A␤ has been poorly examined. Here we report that astrocytes from neonatal rats adhered to surfaces coated with fA␤ and nonfibrillar A␤ (nfA␤). Binding to A␤ was almost completely inhibited by fucoidan and poly(I). Furthermore, we found that SR-MARCO was present both in astrocytes and microglia, and its expression by astrocytes was confirmed on sections of adult rat brain. We propose that SR-MARCO could be responsible for the interaction of astrocytes and microglia with A␤.

EXPERIMENTAL PROCEDURES
Chemicals and Reagents-Fucoidan, poly(C), poly(I), poly-L-lysine, bovine serum albumin (BSA), lidocaine, and trypsin EDTA were purchased from Sigma. Dulbecco's modified Eagle's medium (DMEM), fetal bovine serum, and penicillin/streptomycin were obtained from Invitrogen. Tissue culture dishes and plastic ware were obtained from Falcon (Franklin Lakes, NJ) and Nalgene Nunc (Rochester, NY). The cytotoxicity detection kit (LDH) was obtained from Roche Applied Science. Other reagents were purchased from Sigma or Merck. The A␤-(1-42) was a generous gift from Dr. Heinz Döbeli (Hoffmann-La Roche, Switzerland).
Glial Primary Cultures-Astrocytes and microglial cultures were prepared from brains of 1-2-day-old neonatal rats (37). Newborns were anesthetized with ether and decapitated. Cortices were minced and incubated with 0.25% trypsin-EDTA in Hanks' solution (0.4 g/liter KCl, 0.06 g/liter KH 2 PO 4 , 0.048 g/liter Na 2 HPO 4 , 8 g/liter NaCl, 1 g/liter D-glucose, and 3.5 g/liter NaHCO 3 ) for 15 min and mechanically dissociated. Cells were seeded in 75-cm 2 culture flasks coated with poly-L-lysine (one brain per flask) in complete medium (DMEM/F-12, 10% Fetal Bovine Serum, 100 units/ml penicillin, and 100 g/ml streptomycin) and incubated at 37°C in a water-saturated atmosphere with 5% CO 2 . After 14 -28 days of culture, the flasks were treated with 12 mM lidocaine and shaken at 37°C for 1 h to detach the microglia. Astrocytes were obtained by trypsinization of the astrocytes remaining in the flask. This procedure yielded a highly enriched astrocyte culture (95% or more) and microglial cultures over 99% pure. The cell type of cultures was evaluated by labeling with fluorescein isothiocyanate (FITC)-conjugated lectin Griffonnia simplicifolia (1:200; Sigma), which recognizes microglia, and glial fibrillary acidic protein (GFAP; Dako, Denmark) to identify astrocytes.
Cell Adhesion Assay-A 96-well tissue culture plate was coated with a nonadherent matrix (NAMx), collagen IV for microglia or BSA for astrocytes. Wells were treated with 30 l of 50 g/ml collagen IV or 50 l of 1 mg/ml BSA in distilled water at 37°C for 1 h, washed once with distilled water, and air-dried under sterile conditions. Then ␤-amyloid fibrils (0.1-2.5 g) were added and air-dried. For the adhesion assay, 3 ϫ 10 4 cells/well were plated in 100 l of DMEM/F-12 plus 1 mg/ml BSA in a 96-well plate and placed in an incubator with 5% CO 2 at 37°C for 1 h. The plate was then washed three times with DMEM/ F-12 or PBS. The number of adherent cells was determined by the LDH assay. 3-7 independent experiments were carried out in quadruplicate.
Ligand Competition Assay-Glial cells (5 ϫ 10 4 cells/ml) were preincubated in suspension in DMEM/F-12 plus 1 mg/ml BSA with different known ligands for scavenger receptors for 30 min: 200 g/ml fucoidan (SR-A ligand), 100 g/ml LDL (SR-BI ligand), 100 g/ml oxidized LDL (oxLDL; SR-A and SR-BI ligand), 200 g/ml Mel-BSA (SR-A and SR-B ligand), 200 g/ml poly(I) (SR-A ligand), 200 g/ml poly(C) (negative control for poly(I) adhesion assays). After preincubation, the cells were plated on A␤-coated wells and incubated at 37°C for 1 h. Because preincubation with the competing ligands may result in the endocytosis of cell surface proteins, including receptor complexes, cells were preincubated with competing ligands also at 4 and 20°C to block endocytosis. Wells were washed with PBS, and the number of cells was determined using the LDH assay. Each condition was tested in three independent experiments in quadruplicate.
Binding Inhibition Assay: Immune Neutralization of SR-MARCO-Astrocytes cultures were pretreated for 30 min with or without 20 g/ml rat anti-mouse-SR MARCO IgG (MCA 1849; Serotec) or rat anti-mouse-SR-A IgG (clone 2F8; MCA 1322; Serotec) as control antibody (38) before adding the astrocytes to the 96-well tissue culture plate coated as previously described. After 1 h of adhesion, the cells were washed three times with DMEM/F-12 or PBS. The number of adherent cells was determined by the LDH assay. Three independent experiments were carried out in quadruplicate.
Quantification of Adherent Cells (LDH Assay)-The total content of lactate dehydrogenase (LDH; Roche Molecular Biochemicals) was determined as an indicator of the number of attached cells. Attached cells were also quantified by the incorporation of neutral red, with equivalent results (data not shown). LDH content was chosen because it was more sensitive for discrimination of low cell numbers. In brief, after the cell adhesion assay, culture dishes were extensively washed. Adherent cells were lysed with 100 l of 2% Triton X-100 in PBS. The cell extract was mixed with LDH reaction mixture (1:1). The mixture was incubated in the dark at room temperature for 20 -30 min. The reaction was stopped with 0.2 N HCl. Absorbencies at 490 and 630 nm were measured in a microplate reader. Standard curves were generated by plating 0.5-30 ϫ 10 3 cells and processed in parallel with each LDH assay.
Preparation of Fibrillar and Nonfibrillar ␤-Amyloid-To generate fA␤, A␤-(1-42) was resuspended in Tris buffer and incubated in constant agitation at room temperature for 3-4 days (39,40). The A␤ suspension was centrifuged at 23,000 ϫ g, 4°C for 1 h. Pellet and supernatant correspond to insoluble fA␤ and soluble nfA␤ forms, respectively. Samples were run in a Tris-Tricine SDS-PAGE (41) under reducing and nonreducing conditions. Gels were transferred to a nitrocellulose membrane, blocked, and then probed with the anti-A␤ monoclonal antibody BAP-18 (kindly provided by Dr. H. Döbeli, Hoffmann-La Roche, Switzerland) (42). The blot was then incubated with an antimouse horseradish peroxidase-conjugated secondary antibody. The signal was detected with the chemiluminescence reagent plus kit (PerkinElmer Life Sciences) and captured with a Fuji medical x-ray film. Intensity of the bands was quantified with NIH Image J public domain software available on the World Wide Web at the National Institutes of Health home page.
Congo Red Binding Assay-The assay was established with modifications according to protocols previously established (43). In brief, fA␤ and nfA␤ (1 g) were added to the nonadherent matrix and air-dried. The wells were incubated with a solution of 5 M Congo red (in 20 mM sodium phosphate, 150 mM NaCl, pH 7.4) for 1-24 h at room temperature before the total absorbance was measured at 492 and 540 nm. The amount of Congo red bound to the matrix was used to estimate the quantity of fibrils in the matrix, according to the following: mol of Congo red bound to the matrix ϭ (A 540 nm /25,295) Ϫ (A 492 nm /46,306). As negative control, Congo red binding to nonadherent matrix was also assayed.
Preparation of Native and Oxidized Lipoproteins-LDLs (1.019 -1.063) were isolated by sequential ultracentrifugation of plasma obtained from healthy normolipidemic volunteers anticoagulated with EDTA. Before oxidation, EDTA was removed by dialysis against PBS and sterilized by filtration through a 0.22-m filter. To oxidize LDL, 1 mg/ml lipoprotein was incubated in 5 M CuSO 4 in PBS at 37°C for 24 h. Oxidation was stopped by the addition of 100 M EDTA, followed by dialysis in PBS containing 0.3 mM EDTA for 4 ϫ 24 h. Protein content was determined by the method of Lowry et al. (44). Oxidation was visualized by the increase of the electromobility of oxidized LDL (oxLDL) relative to the native LDL (45).
Indirect Immunofluorescence-Astrocytes, microglia, and mixed glial cultures (50/50) were plated at a density of 2 ϫ 10 4 cells/glass coverslip in 24-well plates. Cells were washed in PBS with 1 mM Ca 2ϩ and fixed in 100% methanol at Ϫ20°C for 5 min. Cells were blocked with 10% goat serum in PBS and then incubated at 4°C overnight with the primary antibodies in blocking solution: polyclonal rabbit anti-GFAP . For the identification of the novel expression of SR-MARCO in astrocytes, two separate antibodies from independent sources were tested: rat anti-mouse MARCO (MCA 1849, 1:100; Serotec) and monoclonal mouse anti-MARCO (1:100; BD Biosciences). Because reports suggested a possible cross-reactivity of the monoclonal mouse anti-MARCO with tubulin, live cells were stained in order to restrict labeling to extracellular epitopes. Except for the samples labeled with lectin, coverslips were washed in PBS and incubated with the corresponding secondary antibody (all from Molecular Probes) diluted 1:100 in blocking solution: Alexa 488-anti-rabbit, Alexa 488anti-mouse, Alexa 546-anti-rabbit, Alexa 546-anti-rat, or Alexa 546anti-goat, at room temperature for 2-4 h. Coverslips were washed in PBS and water. Finally, samples were mounted in fluorescence mounting medium (Dako). As a standard practice, to discard the possibility of nonspecific binding, control assays where the primary antibody was omitted were run in parallel.
Double Immunolabeling on Live Cells-In order to demonstrate cell surface labeling by anti-MARCO, we stained live cells. Cells were plated on glass coverslips as described above. Rat anti-mouse MARCO (MCA 1849; 1:100; Serotec) or monoclonal mouse anti-MARCO (1:100; BD Biosciences) was added to the culture media and incubated at 37°C for 2 h. Cell cultures were washed with PBS plus 0.1 mM Ca 2ϩ /Mg 2ϩ and then fixed with methanol at Ϫ20°C for 30 min. Cells were blocked with 10% goat serum in PBS and then incubated with rabbit anti-GFAP (1:200) or Alexa 488-lectin (1:200) at 4°C overnight. After they were washed, samples were incubated with the secondary antibody: Alexa 546-anti-mouse or Alexa 546-anti-rat (1:100; Molecular Probes) for MARCO and FITC-anti-rabbit (1:100; Molecular Probes) for GFAP at room temperature for 2 h. The samples were rinsed with PBS and mounted in mounting medium (Dako).
Immunohistochemical Staining of Brain Sections-Brains from juvenile (150 g) and adult (250 g) rats were obtained after fixation through cardiac perfusion with 4% p-formaldehyde in phosphate-buffered saline. Those brains were post-fixed in 4% p-formaldehyde for 24 h, and the brains of neonatal rats were fixed for 12 h and then transferred to 15% sucrose at 4°C. Cryosections were prepared at 8 -20 m and stored at Ϫ80°C until use. Immunohistochemistry was performed on attached and free floating sections using the anti-SR-MARCO MCA 1849 antibody. Double staining was carried out using an antibody against GFAP to identify astrocytes. Sections were incubated in blocking solution (PBS supplemented with 10% goat serum, 0.2% Triton X-100, 0.02% azide) for 1 h. Immunodetection was performed by incubation in primary antibodies diluted 1:50 in blocking solution at 4°C for 48 h. Sections were washed four times with PBS, incubated with secondary antibodies diluted 1:100 in blocking solution at room temperature for 2 h, and washed again with PBS. Specific staining was absent if primary antiserum was omitted.
Immunoblot Analysis-Glial cell cultures were prepared as indicated in the description of cell culture methods. Cells were harvested and lysed in an ice-cold lysis buffer (50 mM Tris-HCl, pH 7.5, 150 mM NaCl, 1% Triton X-100, 0.1% SDS, and protease inhibitors). Protein concentration of cell lysates was determined by the BCA method. The lysates (same amount of protein for the different samples of each independent experiment) were dissolved in sample buffer (80 mM Tris-HCl, pH 6.8, 2% SDS, 10% glycerol, 2% 2-mercapthoethanol, 0.02% Coomassie Blue G250) and were heated in a boiling water bath for 5 min. SDS-PAGE was performed with 7.5 and 10% (w/v) polyacrylamide gels and transferred to a nitrocellulose membrane (Amersham Biosciences). After electrophoretic transfer, the membrane was treated with the blocking buffer (PBS plus 0.05% Tween 20, 10% goat serum). The membrane was then incubated with the primary antibody in blocking buffer: anti-SR-A antibody (1:500, Clone 2F8; Serotec), anti-SR-B antibody (1:2000; Novus Biologicals), anti-SR-MARCO antibodies (1:1000, MCA 1849 (Serotec) and 1:1000, clone 22 (BD Bioscience)) or anti-RAGE antibody (1:500; N-16; Santa Cruz Biotechnology). The primary antibody was rinsed and the membrane was incubated with horseradish peroxidaselabeled goat anti-rabbit IgG, horseradish peroxidase-labeled goat antirat IgG, horseradish peroxidase-labeled goat anti-mouse IgG, or horseradish peroxidase-labeled donkey anti-goat IgG in blocking buffer as the secondary antibody, depending on the primary antibody previously used. Signals were detected by enhanced chemiluminescence (Amersham Biosciences) in accordance with the manufacturer's instructions. The molecular mass was estimated with a molecular mass marker kit (Invitrogen). Lysates obtained from adrenal gland (SR-B), liver (SR-A and SR-MARCO), and lung (RAGE) were used as positive controls.
A␤ Removal Assays-Glass slides in 24-wells plates were overlaid with poly-L-lysine and later a central spot of Cy3-A␤ or plain nfA␤ until dried, washed with sterile water, and air-dried. 2 ϫ 10 4 astrocytes were plated on the coated glass slides and incubated for up to 2 days at 37°C. For the experiments with Cy3-A␤, cells were fixed in 5% formalin at room temperature for 25 min, and nuclei were stained with 0.1 g/ml Hoechst 33258 (B2883; Sigma). When unlabeled nfA␤ was used, cells were fixed for 30 min in 4% p-formaldehyde in PBS and permeated with PBS plus 0.2% Triton X-100. Cells were blocked with 5% goat serum in PBS and then incubated with a rabbit polyclonal antibody against glial fibrillary acidic protein (GFAP; 1:200; Dako) as identity marker for astrocytes and a monoclonal antibody anti-A␤ amyloid (BAP-18; 1:200) at 4°C overnight and the corresponding secondary antibody (Molecular Probes): Alexa-546-conjugated anti-mouse (1:100) and Alexa-488-conjugated anti-rabbit (1:100) at room temperature for 2-4 h. Nuclei were also stained with 0.1 g/ml Hoechst 33258.
Statistical Analysis-The statistical significance of the differences in adhesion between the different substrate and competition assays was analyzed with the Mann-Whitney U test. Statistical significance was established for p Ͻ 0.05.

RESULTS
The purity of the cells for the adhesion assays was evaluated by immunocytochemistry with cell identity markers. Microglial cells were over 99% pure, and astrocytes contained less than 5% of contaminating microglia.
Binding of Astrocytes and Microglia of Neonatal Rat to Different Matrices-We compared the adhesion of astrocytes and microglia with different matrices (Fig. 1A). After 1 h of assay, astrocyte adhesion was maximal on poly-L-lysine and minimal on BSA (6 Ϯ 2% cells) (Fig. 1A). Microglial adhesion to poly-L-lysine, plastic, and BSA was 100 Ϯ 3, 116 Ϯ 3, and 91 Ϯ 4%, respectively (Fig. 1A). In contrast, they adhered poorly to gelatin (18 Ϯ 6%) and collagen type IV (29.1 Ϯ 8.6%), as previously reported (46). To determine the adhesion kinetic of astrocytes to different matrices, we plated cells on poly-L-lysine, collagen, and BSA for different times (Fig. 1B). After the first hour, most astrocytes adhered to poly-L-lysine. Adhesion to collagen was poor, and cells did not adhere to BSA during the 24 h of the experiment. Considering these results, BSA was used as the NAMx for astrocytes, and collagen type IV was used for microglia.
Characterization of fA␤ and nfA␤-A␤ aggregation resulted in the formation of soluble nfA␤ and insoluble fA␤ forms. The Western blot analysis ( Fig. 2A) of nfA␤ with nonreducing conditions showed two well defined bands: a 4.1-kDa band corresponding to the monomer (33%) and a second band of 8.4 kDa corresponding to the dimeric form (60%). A third weak band of ϳ12 kDa and a fourth weak band of ϳ36 kDa (likely to be oligomers) were only observed when 1 g of nfA␤ was loaded. Under reducing conditions, two bands were observed at 4.1 kDa (32 and 38%) and at 8.4 kDa (68 and 60%), for 0.5 and 1 g of nfA␤, respectively. A third weak band of ϳ12 kDa was only observed on 1 g of nfA␤. fA␤ under reducing conditions also showed four bands, of 4.1, 8.4, ϳ12, and ϳ36 kDa, that correspond to 43, 54, 3, and 1%, respectively ( Fig. 2A). The immunoblot of fA␤ under nonreducing conditions was not included because the fibers do not enter the electrophoresis gel.
Congo Red Binding Assays-To rule out the aggregation of nfA␤ when the A␤ matrix was prepared, kinetic experiments with an established probe for fibrillar A␤ were performed. The detection of amyloid fibrils in the matrix was followed over time up to 24 h (Fig. 2B). Whereas detection of fA␤ nearly doubled during the 24 h of observation, detection by Congo red in the nfA␤ matrix did not change over time, being only 25% higher than the measurement for the negative control with the nonadhesive matrix. Hence, the results obtained are in agreement with nfA␤ remaining in a nonfibrillar state.
Astrocytes and Microglia of Neonatal Rats Showed Similar Adhesion to Surfaces Coated with nfA␤ and fA␤-Microglia and astrocytes were attached to nonadherent matrix coated with 1 g of fA␤ or nfA␤ after 1 h (Fig. 3A). For quantification, adhesion was compared with the adhesion of glial cells to poly-Llysine. Astrocyte adhesion to surfaces coated with 1 g of fA␤ or nfA␤ was 85.4 Ϯ 6% and 80.1 Ϯ 3.2% (compared with less than 10% for NAMx, p Ͻ 0.001) whether microglial cell adhesion was 72.7 Ϯ 5.3 and 82.8 Ϯ 3.4% (p Ͻ 0.01 compared with NAMx), respectively.
Fucoidan, Poly(I), and oxLDL Block Adhesion of Newborn Rat Astrocytes to nfA␤ and fA␤-Different SR ligands, such as fucoidan, a polysaccharide previously thought to specifically inhibit SR class A type I and II (1), and oxLDL and Mel-BSA (chemically modified lipoprotein and protein, respectively), binding all SRs (47), were tested. In general, most of the ligands that were capable of inhibiting glial adhesion were better competitors for the binding to nfA␤ than for fA␤. Fucoidan inhibited microglial adhesion to fA␤ by 37.5-fold and inhibited adhesion to nfA␤ by 32-fold (p Ͻ 0.001). By comparison, fu- coidan inhibited the adhesion of astrocytes to fA␤ by 5.9-fold and to nfA␤ by 16-fold (Table I; p Ͻ 0.001). Similar results were obtained with poly(I) (also a ligand for type-A SR), which inhibited the adhesion of astrocytes to fA␤ by 7.4-fold and to nfA␤ by 9-fold (p Ͻ 0.01). poly(C) (negative control) did not inhibit the adhesion to fA␤ and only had a mild effect on the adhesion to nfA␤. Native LDL did not inhibit the adhesion of microglia to fA␤ or nfA␤. Similarly, LDL did not significantly affect astrocyte adhesion to fA␤ and only marginally (1.3-fold) reduced the binding of astrocytes to surfaces coated with nfA␤ (Table I). In contrast to its inhibitory effect on the binding of neonatal rat microglia and astrocytes to nfA␤ (a 27-fold reduction in binding, p Ͻ 0.001), oxLDL only reduced partially (6-fold, p Ͻ 0.01) the adhesion of microglia and astrocytes to fA␤ (Table I). Similar results were obtained when cells were preincubated with competing ligands at 20 or 4°C (Table II); poly(I) inhibited the adhesion of astrocytes to fA␤ by 7.4-fold at 37°C, 9.5-fold at 4°C, and 15-fold at 20°C (p Ͻ 0.001 compared with the adhesion in the absence of competing ligand).

Glia of Newborn Rats Expressed Scavenger Receptors and RAGE-
The presence of scavenger receptors and RAGE in astrocytes and microglia were studied by immunocytochemistry (Fig. 4) and immunoblot (Fig. 5). Considering that astrocytes can modulate microglial cells activity and vice versa, the characterization of SRs was performed both on purified cultures of astrocytes and microglial cells and on mixed glial cultures. The immunocytochemistry of mixed glial cultures is shown (Fig. 4), because no differences were found in the expression of any of the SRs tested between glial cells in isolated cultures and in mixed cultures (data not shown). Both microglia (lectin-positive cells in Fig. 4B) and astrocytes (lectinnegative in Fig. 4B) were labeled by an antibody against SR-BI (Fig. 4A). Immunoblot analysis of microglial and astrocytes cultures showed that anti-SR-B antibody recognized two bands at ϳ85 and ϳ60 kDa in both cell types (Fig. 5). In mixed glial cultures, astrocytes (GFAP-positive cells, arrowheads in Fig.  4D) were negative for SR-A, whereas microglia were labeled for SR-AI/AII (small arrows in Fig. 4C). Immunoblot showed that anti-SR-A antibody recognized two bands at ϳ70 and ϳ55 kDa in microglial cell samples (Fig. 5). RAGE also was expressed only by microglial cells (small arrows in Fig. 4E). Immunoblot analysis confirmed that the anti-RAGE antibody labeled two bands at ϳ53 and ϳ60 kDa, present only in microglial cell homogenates (Fig. 5).
The presence of SR-MARCO was evaluated by immunocytochemistry of live cells using two nonrelated antibodies, obtaining similar results. The immunocytochemistry and immunoblot presented in Figs. 4 -6 correspond to those performed with the rat anti-mouse SR-MARCO (MCA 1848 from Serotec). Microglia (lectin-positive; small arrows in Fig. 4H) and astrocytes (GFAP-positive, arrowheads in Fig. 4J) were positive for SR-MARCO (Fig. 4, G and I). Immunoblotting with both anti-SR-MARCO antibodies recognized a single (or a doublet in some samples) band at ϳ54 -55 kDa (Fig. 5).
Immunoblot of cell lysates confirmed the results of the immunocytochemical studies. Astrocytes expressed SR-B and SR-MARCO, whereas no positive bands were detected for SR-A and RAGE. Under the same conditions, immunoblotting of microglial lysates demonstrated positive bands in correspondence to the specific control tissues for the four SRs tested (Fig. 5).
Immune Neutralization of SR-MARCO Blocks Adhesion of Newborn Rat Astrocytes to fA␤-In order to support our data indicating that SR-MARCO is the receptor mediating A␤ adhesion, astrocytes were preincubated with an antibody against SR-MARCO with a known neutralizing effect (38). Immune neutralization of SR-MARCO with the IgG blocked glial adhesion to A␤ by 8-fold (p Ͻ 0.001), similar to the inhibition observed with fucoidan and poly(I) (Fig. 6). In contrast, preincubation with another antibody used as a control, a rat anti-SR-A IgG, reduced glial adhesion only by 18.7 Ϯ 1.6% (p Ͻ 0.001 compared with cells preincubated with anti-SR-MARCO IgG).
Competing Ligands Binding SR-MARCO Inhibit Astrocyte Clearance of A␤ in Vitro-We exposed astrocytes to Cy3-labeled A␤ and nfA␤ in the presence or absence of poly(I), which blocks SR-mediated binding of A␤, or its negative control counterpart, poly(C). After 5 h of co-incubation, phagocytosis was not observed under any of the culture conditions tested (Fig. 7A). After 24 h, whereas control and poly(C)-treated astrocytes showed a clear uptake of A␤, poly(I)-treated astrocytes showed a reduced phagocytosis of Cy3-A␤ (Fig. 7A) and nonlabeled A␤ (Fig. 7B). After 48 h, astrocytes treated with poly(I) phagocytosed Cy3-A␤ poorly, compared with the cell-associated fluorescence observed under control conditions or in cultures exposed to poly(C) (Fig. 7A). Besides the strong astrocyte-associated fluorescence, especially in perinuclear vesicles, the A␤ coating of the matrix was cleared in the area surrounding the astrocytes only by astrocytes that were not exposed to poly(I) (Fig. 7,  A and B).

Scavenger Receptors and Glial Adhesion to A␤
Astrocytes Express SR-MARCO in Brain Sections-To verify the presence of SR-MARCO in brain astrocytes, double immunohistochemistry for SR-MARCO and GFAP was performed in sections of newborn, juvenile, and adult rat brains (Fig. 8). Astrocytes, identified by staining with rabbit anti-GFAP in brains of juvenile (Fig. 8D) and adult (Fig. 8F) rats, stained specifically for SR-MARCO (Fig. 8, C and E, respectively). The cells co-expressing SR-MARCO and GFAP in juvenile and adult rat brains presented the characteristic star-shaped morphology of astrocytes. In contrast, brain sections of newborn rats also showed SR-MARCO staining (Fig. 8A). However SR-MARCOpositive cells had a bi-or unipolar morphology (arrows), and GFAP staining was very pale (Fig. 8B). These observations confirm that astrocytes in rat brains express SR-MARCO.

DISCUSSION
We show that astrocytes from neonatal rats, like microglia, adhered to surfaces coated with both fA␤ and nfA␤. Adhesion to fA␤ and nfA␤ was concentration-dependent for both cell types. Adhesion and uptake of fA␤ by macrophages and microglia is well documented (46,48). It has also been reported that astrocytes from adult but not from neonatal mice bind and degrade fA␤ in culture (36). In contrast, adhesion of astrocytes and microglia to nfA␤ is not well established. Previous reports have only shown that nonfibrillar oligomeric A␤-(1-42) was able to trigger glial activation, which can lead to the production of potentially toxic molecules, such as inflammatory cytokines (interleukin-1␤) and nitric oxide (49). It is interesting that glial cells adhered to nfA␤, because nfA␤ can spontaneously form insoluble assemblies of ␤-pleated sheets (fibrillar) similar to the amyloid fibrils found in AD brains. Binding of nfA␤ to glial cells can lead to the internalization of A␤ through a scavenger receptor. nfA␤ uptake could have two possible consequences: (i) the degradation of internalized nfA␤, which could decrease the formation of fA␤, or (ii) accumulated nfA␤ inside glial cells, which could trigger further aggregation of amyloid fibrils.
Normal subjects and AD patients have similar concentrations of soluble A␤ in the cerebrospinal fluid (50), and A␤ expression by cells in culture is constitutive (51,52). However, the accumulation of amyloid could depend on the rate of degradation of A␤, even in the absence of changes of A␤ production. It has been reported that soluble A␤ is internalized by microglia from newborn mice, through receptors that are not blocked by fucoidan and acetyl-LDL (53). It was suggested that microglia internalized soluble A␤ via pinocytosis, in contrast to receptor-mediated endocytosis of fA␤ (48). In contrast to that report, we found that microglia and astrocyte adhesion to nfA␤ was specifically inhibited by some SR ligands.
Fucoidan and poly(I) similarly inhibited adhesion of both microglia and astrocytes to surfaces coated with fA␤ and nfA␤.
In contrast, oxLDL and Mel-BSA inhibited better the adhesion of astrocytes and microglia to nfA␤ than to fA␤. The different efficacy as competitors of ligands with a wide range of SRspecificity, such as oxLDL and Mel-BSA, suggests that at least some of the receptors involved in the interaction of glial cells with fA␤ and nfA␤ are different. The involvement of different SRs could result in different glial cell activation patterns. The fact that preincubation of cells with competing ligands at 37, 20, and 4°C gave similar results argues against the possibility that the decrease in adhesion to A␤ could be due to the endocytosis of receptor complexes, resulting in decreased cell surface receptor expression. Our results indicate that competing ligands are blocking A␤-SR interaction.
On the other hand, native LDL did not inhibit the adhesion of glial cells to fA␤ or nfA␤. This result suggests that the adhesion of neonatal astrocytes to fA␤ and nfA␤ was mediated by a receptor different from SR-BI (since SR-BI binds native LDL) (54) as previously suggested for astrocytes from adult mice (36). The authors reported that fucoidan and polyinosinic acid inhibited adhesion of astrocytes to A␤, but neutralizing antibodies blocking SR-BI had no effect on A␤ binding. We used 100 g/ml LDL to compete for SR-BI binding, 5-fold the concentration needed to attain maximal binding of COS cells transfected with SR-BI (55). In microglia from an SR-A knockout newborn mouse expressing SR-BI (56), anti-SR-BI only partially blocked whereas fucoidan totally blocked adhesion of SR-A Ϫ/Ϫ microglia to A␤. Our results also suggest that a fucoidan-sensitive receptor other than SR-A and SR-BI could mediate the adhesion of glia to surfaces coated with A␤.
Scavenger-like receptors binding A␤ other than SR-BI are present in astrocytes, including low density lipoprotein receptorrelated protein (5) and membrane-associated proteoglycans. We propose that SR-MARCO, an SR closely related to SR-AI (33,57), that we immunodetected in astrocytes and microglia could bind A␤. This proposition is strongly supported by the fact that a neutralizing antibody for SR-MARCO prevented adhesion of glial cells to A␤. The magnitude of the adhesion inhibition was similar to that observed with poly(I) and fucoidan. In contrast, anti-SR-AI/II IgG had no effect on adhesion of astrocytes to A␤. These findings lead us to conclude that SR-MARCO is probably the member of the SR-A family promoting astrocyte adhesion to A␤.
We found that neonatal astrocytes expressed SR-BI and SR-MARCO but not SR-AI/AII or RAGE. In contrast, microglia presented all of them. Previous reports showed that murine neonatal microglia in culture express SR-A, SR-BI, and CD36 (48), whereas adult mouse microglia only express low levels of CD36 (2) and do not express detectable levels of SR-A (58). On the other hand, the expression of receptors also changes under certain pathophysiological conditions. SR-A is expressed by activated microglia in the vicinity of A␤-containing senile plaques (59,60).
Astrocytes from adult mouse and human brains and from AD brains express SR-BI (15). We presented the first description of the presence of SR-MARCO and SR-BI in astrocytes from neonatal rat in culture. However, the extrapolation of results obtained with microglia or astrocytes from neonatal animals to the adult condition must be done with caution, because the pattern of SR expression changes with age (2,36). We also confirmed SR-MARCO expression in the brains of rats from neonatal to adult stages. Astrocytes in the juvenile and adult rat express SR-MARCO. The presence of SR-MARCO in the brain raises new questions relevant to brain cell physiology in health and disease. In brain slices, the astrocytes expressing SR-MARCO were mostly GFAP-positive. This pattern of expression suggests that SR-MARCO was expressed mainly by activated astrocytes. The restricted pattern of expression of SR-MARCO is not contradictive with the expression of SR-MARCO by astrocytes in culture. In fact, there is a deregulation of GFAP expression in astrocytes in culture, and nearly all astrocytes are GFAP-positive. An expression of SR-MARCO restricted to activated astrocytes could have important consequences for the proposition that it may be involved in the interaction of A␤ with glial cells in Alzheimer disease. The more restricted expression of SR-MARCO relative to the other SR-As in normal tissue suggests that its expression by astrocytes in the human brain and on astrocytes surrounding amyloid plaques will be worth examination.

FIG. 7. SR-A-competing ligands inhibit phagocytosis and clearance of A␤ by astrocytes in vitro.
Shown is fluorescence microscopy of astrocytes plated on a glass slide coated with Cy3-labeled A␤ (A) or unlabeled nfA␤ (B) as described under "Experimental Procedures." The A␤ removal assay was performed with astrocytes under control conditions or exposed to poly(I) or poly(C). A, double fluorescence microscopy of neonatal rat astrocytes plated on poly-L-lysine plus Cy3-labeled A␤-coated surfaces for 5, 24, and 48 h. In the merged image, red corresponds to the Cy3-labeled A␤, and blue corresponds to the Hoechst-stained nucleus. B, neonatal rat astrocytes plated on poly-L-lysine plus unlabeled nfA␤-coated surfaces for 24 h. Cells were double immunolabeled with an antibody recognizing GFAP (as an identity marker for astrocytes, in green) and an antibody against A␤ (in red). Hoechst-stained nuclei are shown in blue. Merged images allow viewing of the uptake of A␤ by the astrocytes, a process greatly reduced in cells exposed to poly(I).
The possible functional role for astrocyte adhesion to A␤ is another issue. Whereas it has been shown that microglia bind to and internalize A␤ and can degrade it under certain conditions, little is known regarding such activity by astrocytes. We found that neonatal rat astrocytes phagocytosed and cleared surface-bound A␤, phagocytosis that was partially inhibited by poly(I). These observations, the expression of SR-MARCO by GFAP-positive astrocytes, and the presence of A␤ in astrocytes surrounding amyloid plaques (61)(62)(63)(64) suggest that astrocytes could participate in the degradation of A␤ in AD or that dysregulation of A␤ clearance by astrocytes may be responsible for its accumulation in AD. A rapid clearance of apoptotic cells, oxidized compound, and other typical SR ligands from the central nervous system is needed to prevent further damage and disease. In the phagocytosis assay, also the astrocytes exposed to poly(I) phagocytosed A␤ after 48 h. However, the phagocytosis was slower, and the astrocytes were unable to clear the surface-bound A␤ as was observed for control and poly(C)-treated astrocytes. The partial A␤ uptake could depend on the up-regulation of other SR. It is necessary to consider that the poly(I) used to block A␤ uptake binds to the SR and can mediate cell activation. Astrocytes could interact with A␤ without necessarily performing phagocytosis. Astrocytes become rapidly activated in response to various injuries. In many diseases, astrocyte activation has been associated with induction of chemokines and inflammatory cytokines.
An aspect that deserves a closer look is the activation of signaling pathways associated with the binding of SR. We speculate that although A␤ binds to several different SRs, differences in cell response could depend on the activation of specific intracellular signaling pathways. Although there is no information regarding SR-MARCO, ligand binding to related SRs like SR-A stimulates activation of phosphatidylinositol 3-kinase and protein-tyrosine kinase and mitogen-activated protein pathways (65,66), and protein kinase C inhibitors reduced myelin phagocytosis mediated by SR-A and complement receptor (67). Binding to SR-A also stimulates increases in proinflammatory cytokines, such as interleukin-1␤ and tumor necrosis factor ␣ (65, 66), related to neurotoxicity in degenerative diseases. It has been well established that astrocytes regulate microglial activation by A␤ and proinflammatory molecules, including the production of reactive oxygen species (63,64,68), and they also influence the interaction of microglia with A␤, changing the phagocytic capability of microglial cells (69) and microglial cell-mediated neurotoxicity (70). However, it is still unclear whether reactive astrocytes accumulating at sites of A␤ deposition have neuroprotective or damaging functions (71). Although astrocytes do not produce reactive oxygen species (15,56), they produce important amounts of nitric oxide when they adhere to matrices containing A␤. Astrocytes and microglial cells could also influence their expression pattern for different SRs. However, at least under our culture conditions, we did not find differences in the expression of SR-A, SR-B, SR-MARCO, or RAGE between astrocytes and microglia in isolated cultures and those in mixed cultures.
In summary, considering the findings regarding the adhesion pattern of astrocytes and microglia to fA␤ and nfA␤ (poorly examined by previous work), the fact that binding of neonatal rat microglia and astrocytes is mediated via fucoidan-and poly(I)-sensitive mechanisms, and the expression of SR-MARCO in their surface both in culture and in the rat brain, we propose that SR-MARCO could be the scavenger receptor responsible for the uptake of A␤ by astrocytes and microglia. We are presently evaluating if the activation of microglia or astrocytes by proinflammatory cytokines or A␤ induces the up-regulation of SR-MARCO. We are also studying the activation of specific signal transduction pathways in response to A␤ binding.
FIG. 8. Immunostaining of brain sections from neonatal, juvenile, and adult rats for SR-MARCO. Neonatal (A and B), juvenile (C and D), and adult (E and F) brain (cortex) double stained with rat anti-SR-MARCO and Alexa 546-conjugated goat anti-rat IgG (A, C, and E) and rabbit anti-GFAP and Alexa 488-conjugated goat anti-rabbit IgG (B, D, and F). In newborn rat brains, SR-MARCO-positive cells were not star-shaped, but they had polarized long processes (small arrows in A) and show a faint labeling for GFAP (B). In contrast, in juvenile and adult animals, double stained cells are clearly identified as astrocytes by their star-shaped morphology (arrowheads in C and E) and correspond to GFAP-positive cells (arrowheads in D and F). Some GFAPpositive cells were not labeled by anti-SR-MARCO (*). A few round and elongate shaped cells labeled by anti-SR-Marco (small arrows in C) and negative for GFAP labeling (small arrows in D) were observed in the brain sections of juvenile rats. Scale bar, 50 m.