Adult mouse astrocytes degrade amyloid- in vitro and in situ

Alzheimer disease (AD) is a progressive neurodegenerative disorder characterized by excessive deposition of amyloid- (A) peptides in the brain. One of the earliest neuropathological changes in AD is the accumulation of astrocytes at sites of A deposition¹, but the cause or significance of this cellular response is unclear. Here we show that cultured adult mouse astrocytes migrate in response to monocyte chemoattractant protein-1 (MCP-1), a chemokine present in AD lesions¹, and cease migration upon interaction with immobilized A₁₋₄₂. We also show that astrocytes bind and degrade A₁₋₄₂. Astrocytes plated on A-laden brain sections from a mouse model of AD associate with the A deposits and reduce overall A levels in these sections. Our results suggest a novel mechanism for the accumulation of astrocytes around A deposits, indicate a direct role for astrocytes in degradation of A and implicate deficits in astroglial clearance of A in the pathogenesis of AD. Treatments that increase removal of A by astrocytes may therefore be a critical mechanism to reduce the neurodegeneration associated with AD.

The presence of large numbers of astrocytes associated with A deposits in AD suggests that these lesions generate chemotactic molecules that mediate astrocyte recruitment. In fact, AD lesions contain MCP-1 (ref. 1), a potent chemoattractant for neonatal astrocytes in vitro², but the cellular source of MCP-1 in the AD brain is unclear. Whereas neonatal astrocytes from various species produce MCP-1 after stimulation with A₁₋₄₂ in vitro^{3, 4, 5}, astrocytes surrounding A plaques in the AD brain and in a mouse model for AD do not express detectable levels of MCP-1 (refs. 6, 7). These findings prompted us to compare A₁₋₄₂-stimulated release of MCP-1 by astrocytes cultured from neonatal and adult mouse brains. We found that A₁₋₄₂-stimulated neonatal mouse astrocytes produced significantly (P = 0.05) higher amounts of MCP-1 compared with non-stimulated controls, whereas similarly stimulated adult astrocytes do not (Table 1). However, both neonatal and adult astrocytes significantly (P = 0.0001 and 0.01, respectively) increased MCP-1 release in response to lipopolysaccharide (LPS; Table 1). Because cultured adult mouse astrocytes seem to be refractory to A-induced MCP-1 production, similar to astrocytes surrounding A deposits in AD, we used these cells for most of our experiments.

Table 1. MCP-1 production (pg/ml) by neonatal or adult mouse astrocytes after stimulation with A1−42 or LPS. Full Table

We used cell culture inserts to test the ability of chemoattractants MCP-1 and LPS⁸ to stimulate migration of adult astrocytes across a porous membrane coated with collagen IV (CIV), an extracellular matrix protein found in A plaques (Fig. 1a). In the presence of 10^-8 M MCP-1, 2-fold more astrocytes migrated across CIV-coated membranes than in the absence of MCP-1. Control experiments with LPS produced similar results (Fig. 1a). Migration of astrocytes in response to either chemoattractant was reduced to non-stimulated levels when the membranes were coated with CIV and A₁₋₄₂ (Fig. 1a). These studies indicate that astrocytes can be recruited to sites of A deposition by locally released MCP-1 or other chemoattractants, and become immobilized when they contact A in the extracellular matrix.

Figure 1. Adult mouse astrocytes cease migration upon interaction with A1−42 and adhere to A1−42-coated surfaces. a, Migration of astrocytes, in response to LPS or MCP-1, across porous membranes coated with CIV alone or with CIV and A1−42. Number of cells that migrated across CIV-coated membranes under unstimulated conditions (214 74 cells/mm2; n = 4) was designated as 100%. #, P = 0.05 compared with no chemoattractant; †, P = 0.01 compared with LPS and A1−42; *, P = 0.05 compared with MCP-1 and A1−42. b, Adhesion of astrocytes, suspended in buffer with () or without () Ca2+ and Mg2+, to multi-spot glass slides that were uncoated, coated with CIV or coated with CIV and A1−42 (2 g/spot). c, Adhesion of astrocytes, suspended in buffer lacking Ca2+ and Mg2+, to multi-spot glass slides coated with CIV and increasing amounts of A1−42. d, Binding of Cy3-labeled A1−42 (Cy3-A1−42) to astrocytes with or without addition of 100 or 500 g/ml fucoidan, polyinosinic acid (polyI), or antibody against SRBI (SRBI). Values are mean s.e.m.; n = 3−5 experiments. *, P = 0.05 compared with basal; **, P = 0.01 compared with basal. Full Figure and legend (32K)

Immobilization of adult astrocytes upon interaction with A indicates that these cells adhere to A-coated surfaces. To examine astrocyte adhesion, we took advantage of the fact that adhesion of astrocytes to CIV-coated surfaces is mediated by integrins⁹ and is therefore dependent on divalent cations. In the absence of Ca²⁺ and Mg²⁺, astrocyte adhesion to CIV-coated multi-spot glass slides was <3% of that in the presence of Ca²⁺ and Mg²⁺ (Fig. 1b). Additional coating with human or mouse A₁₋₄₂ promoted astrocyte adhesion even in the absence of Ca²⁺ and Mg²⁺ (Fig. 1b and data not shown). Over 90% of added cells adhered to spots coated with 2 g mouse or human A in a dose-dependent fashion (Fig. 1c and data not shown). These findings suggest that astrocytes associate with A in the extracellular matrix through Ca²⁺- and Mg²⁺-independent receptor(s). One candidate for such a receptor is scavenger receptor class B type I (SRBI), an A-binding receptor expressed on astrocytes (reviewed in ref. 10). We exposed astrocytes to Cy3-labeled A₁₋₄₂ (Cy3-A₁₋₄₂) in the presence or absence of fucoidan or polyinosinic acid, substances that block scavenger receptor−mediated binding of A₁₋₄₂ by a variety of cells¹⁰. Cy3-A₁₋₄₂−treated astrocytes displayed a strong cell-associated fluorescence that was reduced substantially by fucoidan and polyinosinic acid in a concentration-dependent manner (Fig. 1d). However, antibodies blocking A-SRBI interaction¹¹ had no effect on A binding to cells (Fig. 1d). These results indicate that A binds to astrocytes through molecules that share characteristics with scavenger receptors. In fact, other known A-binding molecules present on astrocytes, including the receptor for advanced glycation end products (RAGE), low-density lipoprotein receptor−related protein or membrane-associated proteoglycans, may synergize with scavenger receptor−like molecules in binding A, as previously shown for microglia¹¹.

These observations and the presence of A in astrocytes surrounding amyloid plaques^{12, 13, 14, 15} indicated that astrocytes not only adhere to and are immobilized by A, but also degrade it. To test this hypothesis, we examined surfaces coated with Cy3-A₁₋₄₂ by fluorescence microscopy before (Fig. 2a) and after incubation with adult astrocytes for 24−48 h (Fig. 2b−d). After 24 h, astrocytes were well spread (Fig. 2b) and Cy3-A₁₋₄₂ had disappeared from the surfaces underneath and surrounding the cells but was present in perinuclear vesicles (Fig. 2c). After 48 h, perinuclear fluorescence disappeared almost completely (Fig. 2d). Similar observations were made using surfaces coated with unlabeled A₁₋₄₂ followed by immunohistochemical detection of intracellular A with an A-specific antibody (data not shown). In contrast, astrocytes cultured from neonatal brain were much less efficient at removing immobilized Cy3-A₁₋₄₂ (Fig. 2e and f). These findings show that, in contrast to cultured neonatal mouse or rat astrocytes¹⁶, adult mouse astrocytes clear immobilized A.

Figure 2. Adult mouse astrocytes clear A1−42in vitro and in situ. a, Fluorescence microscopy of multi-spot glass slide coated as previously described23 with Cy3-labeled A1−42. b−d, Adult mouse astrocytes plated on Cy3-A1−42−coated surfaces for 24 h (b and c) and 48 h (d), imaged by phase contrast (b) or fluorescence (c and d) microscopy. e and f, Neonatal mouse astrocytes plated on Cy3-A1−42−coated surfaces for 24 h, imaged by phase contrast (e) and fluorescence (f) microscopy. c, d, f, Red, Cy3-A1−42 signal; blue, DAPI-stained nuclei (arrows). g−i, Reduction of A-immunoreactive area in the hippocampus of human APP−expressing mouse brain sections. Representative sections incubated without (g) or with (h) adult astrocytes for 24 h and immunostained for A with 3D6 antibody. Quantitation from 1 of 4 similar experiments (i) shows significantly less A immunoreactivity in sections incubated with astrocytes (n = 9 sections) compared with sections without astrocytes (n = 10 sections). *, P = 0.05. j−l, Confocal images from a human APP−transgenic brain section incubated with adult astrocytes. An astrocyte immunostained for S100 (green) adheres to the section (j). A-immunopositive material (red) stained with A-specific antibody 1280 (k) seems closely associated with the astrocyte in the merged image (l). Scale bar, 100 m. Full Figure and legend (88K)

To determine whether adult astrocytes could also process A deposits from brain tissue, we plated cells on unfixed A-rich brain sections from transgenic mice expressing human amyloid precursor protein (APP). After incubation of brain sections with astrocytes for 24 h, the total hippocampal area occupied by A was up to 40% lower than in untreated sections (Fig. 2g−i). The astrocytes were spread out over the tissue (Fig. 2j) and many were intimately associated with A-immunoreactive material, as shown by confocal microscopy (Fig. 2j−l). No A immunoreactivity was observed in astrocytes cultured on brain sections from non-transgenic mice (data not shown). These results indicate that astrocytes remove A deposits from brain sections in situ.

To quantitate the degradation of A by adult astrocytes, we exposed these cells to A₁₋₄₂ and determined the A content of the cell fraction and cell-culture supernatant by western blot analysis. Within 3 h, A₁₋₄₂ levels decreased in the supernatant and increased in the cell-associated fraction, indicating that astrocytes removed A from the medium. After 3 h, A levels in the cell fraction began to decrease, and by 24−48 h there was virtually no A detectable in the supernatant or the cell fraction (Fig. 3a). These findings show that cultured adult astrocytes degrade A₁₋₄₂. Similar results were obtained using standard ELISA procedures to measure A₁₋₄₂ levels in the same cell cultures (Fig. 3b).

Figure 3. Degradation of A1−42 by adult mouse astrocytes as measured by western blot and ELISA. a and b, Astrocytes were incubated with synthetic A1−42, and supernatant or adherent cells (pellet) were collected after the indicated time points. Monomeric A is detected in the cell pellet (a, left panel; b, ) after 3 h and disappears at 48 h in both pellet and supernatant (a, right panel; b, ). *, non-specific band. Higher molecular weight bands in pellet and supernatant may be A polymers or aggregates with other proteins that are also degraded. One representative experiment is shown (n = 3 experiments). Full Figure and legend (61K)

Astrogliosis is one of the earliest pathological manifestations of AD and may be a response to the increasing number of degenerating neurons and synapses, or to the accumulation of A. Astrocyte activation seems to be particularly prominent around A deposits both in the brain parenchyma and in the cerebrovasculature. It is possible that astrogliosis in AD occurs as a response to the MCP-1 present in these A deposits¹. We found that adult astrocytes, in contrast to neonatal astrocytes^{3, 4, 5}, do not respond to stimulation with A by increasing their release of MCP-1 (Table 1). This suggests that the MCP-1 found in AD lesions is of microglial, but not astrocytic origin⁶. In addition, our results show that A associated with extracellular matrix proteins such as CIV can bind to and directly immobilize migrating astrocytes (Fig. 1), a process that may contribute to the accumulation of astrocytes around A deposits in the AD brain.

It is unclear whether 'reactive' astrocytes accumulating at sites of A deposition have neuroprotective or destructive functions (for review, see ref. 17). Whereas some studies implicate neonatal rat astrocytes in the clearance of A16, others show that neonatal mouse astrocytes inhibit microglial phagocytosis of senile plaque cores isolated from AD brains¹⁸. Our current findings show that in contrast to neonatal mouse and rat astrocytes¹⁶, adult mouse astrocytes efficiently clear exogenously added and surface-bound A (Figs. 2 and 3) and are capable of removing A deposited in brain slices from mice expressing human APP (Fig. 2). These data and those in Table 1 also show that cultured adult astrocytes differ qualitatively from neonatal astrocytes. Thus, functional and phenotypic differences between cultured neonatal and adult astrocytes must be considered when these cells are used to model astrocyte functions in normal adult brain and in diseases such as AD.

In our paradigm, exogenous adult astrocytes were able to remove and degrade A without additional stimuli such as opsonins or cytokines, whereas microglia seem to require stimulation by cytokines or the presence of opsonizing antibodies¹⁹. Our results are surprising because endogenous astrocytes surround and contact A deposits in AD brain but seem incapable of removing A. They raise the possibility that dysregulation of A clearance by astrocytes may precede or be responsible for the accumulation of A in AD. Further studies are needed to determine whether astrocytes degrade A₁₋₄₂ intra- or extracellularly, and whether A clearance from the brain is among the normal homeostatic functions of astrocytes. Because astrocytes greatly outnumber microglia in the brain²⁰, these cells could have a more critical role in A removal than previously thought. Although numerous studies show that excessive astrocyte activation can promote neurodegeneration, increasing astrocyte-mediated clearance of A may be a strategy for reducing A-associated toxicity and progression of AD.

Astrocytes were purified as described^{4, 21}, with the following modifications. Cortices of neonatal (1−3 d) and adult (6−8 weeks) C57BL/6 mice were digested in Ca²⁺- and Mg²⁺-free HBSS with 0.25% trypsin and 1 mM EDTA (and 0.04 mg/ml bovine pancreatic DNase I for adult cortices), for 30 min at 37 °C in tissue culture flasks on a rotary shaker at 150 r.p.m. (4 neonatal or 2 adult hemispheres per 10ml per flask). Cells were washed and replated in DMEM/F12 complete (DMEM/F12 containing 10% heat-inactivated FBS, penicillin and streptomycin; Gibco, Rockville, Maryland) to generate primary cultures. Astrocyte purity was >99%, as determined by immunofluorescence using antibodies specific for glial fibrillary acidic protein (Santa Cruz Biotechnology, Santa Cruz, California) and S100 (Sigma, St. Louis, Missouri). Antibodies specific to CD11b (Serotec, Raleigh, North Carolina) and galactocerebroside (Sigma) were used to identify microglia or oligodendrocytes, respectively.

A.

Human (Bachem, Torrance, California) and mouse (Anaspec, San Jose, California) A₁₋₄₂ were dissolved in double-distilled water at 1 mg/ml and incubated at 37 °C (for details, see ref. 22). Human Cy3-A₁₋₄₂ was prepared as described²³.

Astrocytes in 100 l DMEM/F12 containing N₂ supplements, penicillin and streptomycin were plated into 96-well plates at 5 10⁴ cells/well. After 24 h, cultures were incubated in 100 l fresh medium with or without 25 M human A₁₋₄₂ or 10 g/ml LPS (Sigma) for an additional 24 h. Mouse MCP-1 in the supernatants was measured by ELISA (Pharmingen, San Diego, California).

Multi-spot glass slides (Shandon, Pittsburgh, Pennsylvania) were coated with CIV (Fluka, Milwaukee, Wisconsin; 50 g/ml in water for 1 h), air-dried and overlaid with various amounts of human or mouse A₁₋₄₂, as described¹¹. Astrocytes (5 10⁴ cells/spot) suspended in HBSS with or without Ca²⁺ and Mg²⁺ were plated on peptide-coated spots, incubated for 45 min at 37 °C and washed. Adherent cells were quantitated using CyQuantGR (Molecular Probes, Eugene, Oregon) as described¹¹.

Cell culture inserts (8-m pore size; Becton Dickinson, Franklin Lakes, New Jersey) were coated with CIV and overlaid with 8 g human A₁₋₄₂. Astrocytes (10⁵ cells) suspended in RPMI 1640 with 25 mM HEPES and 0.1% BSA were added to the upper chamber of the inserts; LPS (100 g/ml) or MCP-1 (10^-8 M; Leinco Technologies, St. Louis, Missouri) was added to the bottom chamber. After incubation at 37 °C for 6 h, non-migrated cells on the upper surface of the membrane were removed by scraping. Cells that migrated to the lower surface of the membrane were stained with hematoxylin and counted with an inverted microscope.

A degradation.

Astrocytes (10⁶ cells/well) in serum-free neurobasal medium containing N₂ supplement and 2 mM L-glutamine were exposed to 2 g/ml human A₁₋₄₂. At various time points, A levels in cell-culture supernatants and adherent cells were determined by immunoblotting as described²⁴, with antibody 266 (against human A₁₃₋₂₈) or 3D6 (against human A₁₋₅; 2 g/ml; P. Seubert, Elan Pharmaceuticals, South San Francisco, California). A₁₋₄₂ levels were also measured in culture supernatants and cell fractions homogenized in 5 M guanidine buffer, using ELISA with capture antibody 21F12 (against human A₃₃₋₄₂) and biotinylated reporter antibody 3D6 as described²⁴.

Astrocyte cultures maintained in Krebs-Ringer buffer containing 1 mM glucose and 0.1% bovine serum albumin (KRBGA) were pretreated for 30 min at 37 °C with or without fucoidan, polyinosinic acid (100 and 500 g/ml) or 200 g/ml SRBI-specific rabbit antibody (Novus Biologicals, Littleton, Colorado), before adding 10 g/ml Cy3-A₁₋₄₂ in KRBGA as described¹¹. After 1 h, cells were washed with KRBGA, fixed in 5% formalin and photographed. Cy3 fluorescence intensity was quantified using NIH Image software (NIH, freeware at http://rsb.info.nih.gov/nih-image).

A removal.

Multi-spot glass slides were overlaid with DMEM/F12 containing 10 g/ml Cy3-A₁₋₄₂ for 1 h at 37 °C, washed with sterile water and air-dried. Astrocytes suspended in DMEM/F12 complete were plated on Cy3-A₁₋₄₂−coated spots (5 10³ cells per 50 l per spot) and incubated for up to 2 d at 37 °C. Cells were fixed in 5% formalin, nuclei were stained with DAPI, and fluorescence and phase images were taken.

Clearance of A from APP mouse brain slices.

Brains of saline-perfused 22-month-old mice expressing human APP under control of the platelet-derived growth factor B chain promoter²⁵ (line J20; L. Mucke, Gladstone Institute of Neurological Disease) were divided sagittally and snap frozen. Cryosections (10 m) were mounted on poly-L-lysine-coated coverslips, transferred to 12-well plates and incubated with or without adult astrocytes (6 10⁵ /well) for 24 h at 37 °C. Sections were then fixed in 4% paraformaldehyde, treated with 0.1% Triton X-100, immunostained with 3D6 antibody (1:1000) and developed with diaminobenzidine and hydrogen peroxide. Relative areas of the hippocampus occupied by 3D6 immunoreactivity were measured for 8−15 sections per condition with Bioquant 98 software (R&M Biometrics, Nashville, Tennessee). For confocal microscopy, sections similarly incubated with astrocytes were fixed in acetone and immunolabeled with rabbit antibody 1280 against human A (1:1000; D. Selkoe, Harvard University, Boston, Massachusetts) and goat antibody against S100 (1:1000). After washing, sections were incubated with FITC- or Texas Red−conjugated secondary antibodies (Vector Laboratories, Burlingame, California), mounted and imaged by scanning confocal microscopy. All animal experiments were performed in accordance with institutional guidelines.

	Top

1. Role of inflammatory processes and microglial activation in Alzheimer's disease. In Alzheimer's Disease (ed. Terry, R.D.) chap. 26, 390 (Lippincott Williams & Wilkins, Philadelphia, Pennsylvania, 1999).
2. Heesen, M. et al. Mouse astrocytes respond to the chemokines MCP-1 and KC, but reverse transcriptase-polymerase chain reaction does not detect mRNA for the KC or new MCP-1 receptor. J. Neurosci. Res. 45, 382–391 (1996). | Article | PubMed  | ISI | ChemPort |
3. Hayashi, M., Luo, Y., Laning, J., Strieter, R.M. & Dorf, M.E. Production and function of monocyte chemoattractant protein-1 and other -chemokines in murine glial cells. J. Neuroimmunol. 60, 143–150 (1995). | Article | PubMed  | ISI | ChemPort |
4. Johnstone, M., Gearing, A.J. & Miller, K.M. A central role for astrocytes in the inflammatory response to -amyloid; chemokines, cytokines and reactive oxygen species are produced. J. Neuroimmunol. 93, 182–193 (1999). | Article | PubMed  | ISI | ChemPort |
5. Rezaie, P., Trillo-Pazos, G., Everall, I.P. & Male, D.K. Expression of -chemokines and chemokine receptors in human fetal astrocyte and microglial co-cultures: potential role of chemokines in the developing CNS. Glia 37, 64–75 (2002). | Article | PubMed  | ISI | ChemPort |
6. Ishizuka, K. et al. Identification of monocyte chemoattractant protein-1 in senile plaques and reactive microglia of Alzheimer's disease. Psychiat. Clin. Neurosci. 51, 135–138 (1997). | ISI | ChemPort |
7. Mehlhorn, G., Hollborn, M. & Schliebs, R. Induction of cytokines in glial cells surrounding cortical -amyloid plaques in transgenic Tg2576 mice with Alzheimer pathology. Int. J. Dev. Neurosci. 18, 423–431 (2000). | Article | PubMed  | ISI | ChemPort |
8. Tezel, G., Hernandez, M.R. & Wax, M.B. In vitro evaluation of reactive astrocyte migration, a component of tissue remodeling in glaucomatous optic nerve head. Glia 34, 178–189 (2001). | Article | PubMed  | ISI | ChemPort |
9. Tawil, N.J. et al. 1 1 integrin heterodimer functions as a dual laminin/collagen receptor in neural cells. Biochemistry 29, 6540–6544 (1990). | PubMed  | ISI | ChemPort |
10. Husemann, J., Loike, J.D., Anankov, R., Febbraio, M. & Silverstein, S.C. Scavenger receptors in neurobiology and neuropathology: their role on microglia and other cells of the nervous system. Glia 40, 195–205 (2002). | Article | PubMed  | ISI |
11. Husemann, J., Loike, J.D., Kodama, T. & Silverstein, S.C. Scavenger receptor class B type I (SR-BI) mediates adhesion of neonatal murine microglia to fibrillar -amyloid. J. Neuroimmunol. 114, 142–150 (2001). | Article | PubMed  | ISI | ChemPort |
12. Wegiel, J., Wang, K.C., Tarnawski, M. & Lach, B. Microglia cells are the driving force in fibrillar plaque formation, whereas astrocytes are a leading factor in plague degradation. Acta Neuropathol. (Berl.) 100, 356–364 (2000). | Article | PubMed  | ISI | ChemPort |
13. Kurt, M.A., Davies, D.C. & Kidd, M. -amyloid immunoreactivity in astrocytes in Alzheimer's disease brain biopsies: an electron microscope study. Exp. Neurol. 158, 221–228 (1999). | Article | PubMed  | ISI | ChemPort |
14. Thal, D.R. et al. Amyloid -protein (A)-containing astrocytes are located preferentially near N-terminal-truncated A deposits in the human entorhinal cortex. Acta Neuropathol. (Berl.) 100, 608–617 (2000). | Article | PubMed  | ISI | ChemPort |
15. Yamaguchi, H., Sugihara, S., Ogawa, A., Saido, T.C. & Ihara, Y. Diffuse plaques associated with astroglial amyloid protein, possibly showing a disappearing stage of senile plaques. Acta Neuropathol. (Berl.) 95, 217–222 (1998). | Article | PubMed  | ISI | ChemPort |
16. Shaffer, L.M. et al. Amyloid protein (A ) removal by neuroglial cells in culture. Neurobiol. Aging 16, 737–745 (1995). | Article | PubMed  | ISI | ChemPort |
17. Wyss-Coray, T. & Mucke, L. Inflammation in neurodegenerative disease—a double-edged sword. Neuron 35, 419–432 (2002). | Article | PubMed  | ISI | ChemPort |
18. DeWitt, D.A., Perry, G., Cohen, M., Doller, C. & Silver, J. Astrocytes regulate microglial phagocytosis of senile plaque cores of Alzheimer's disease. Exp. Neurol. 149, 329–340 (1998). | Article | PubMed  | ISI | ChemPort |
19. Bard, F. et al. Peripherally administered antibodies against amyloid -peptide enter the central nervous system and reduce pathology in a mouse model of Alzheimer disease. Nat. Med. 6, 916–919 (2000). | Article | PubMed  | ISI | ChemPort |
20. Savchenko, V.L., McKanna, J.A., Nikonenko, I.R. & Skibo, G.G. Microglia and astrocytes in the adult rat brain: comparative immunocytochemical analysis demonstrates the efficacy of lipocortin 1 immunoreactivity. Neuroscience 96, 195–203 (2000). | Article | PubMed  | ISI | ChemPort |
21. De Groot, C.J. et al. Establishment of human adult astrocyte cultures derived from postmortem multiple sclerosis and control brain and spinal cord regions: immunophenotypical and functional characterization. J. Neurosci. Res. 49, 342–354 (1997). | Article | PubMed  | ChemPort |
22. Coraci, I.S. et al. CD36, a class B scavenger receptor, is expressed on microglia in Alzheimer's disease brains and can mediate production of reactive oxygen species in response to -amyloid fibrils. Am. J. Pathol. 160, 101–112 (2002). | PubMed  | ISI | ChemPort |
23. Paresce, D.M., Ghosh, R.N. & Maxfield, F.R. Microglial cells internalize aggregates of the Alzheimer's disease amyloid -protein via a scavenger receptor. Neuron 17, 553–565 (1996). | Article | PubMed  | ISI | ChemPort |
24. Wyss-Coray, T. et al. TGF-1 promotes microglial amyloid- clearance and reduces plaque burden in transgenic mice. Nat. Med. 7, 612–618 (2001). | Article | PubMed  | ISI | ChemPort |
25. Mucke, L. et al. High-level neuronal expression of a_1–42 in wild-type human amyloid protein precursor transgenic mice: synaptotoxicity without plaque formation. J. Neurosci. 20, 4050–4058 (2000). | PubMed  | ISI | ChemPort |

	Top

Competing interests statement:  The authors declare that they have no competing financial interests.