Beta-amyloid peptide secretion by a microglial cell line is induced by beta-amyloid-(25-35) and lipopolysaccharide.

β-Amyloid protein (βAP) deposition is a neuropathologic hallmark of Alzheimer's disease (AD). Yet, the source of cerebral βAP in AD is controversial. We examined the production of βAP by the BV-2 immortalized microglial cell line using a sensitive enzyme immunoassay. Constitutive production of βAP was detected in conditioned media from unstimulated BV-2 cells. Further, production of βAP was induced by treatment of cultures by lipopolysaccharide (LPS) or βAP-(25–35) and was inhibited by the calpain protease inhibitor MDL 28170. Treatment of BV-2 cells with LPS or βAP-(25–35) did not affect cell-associated β-amyloid precursor protein levels. These findings suggest that microglia may be an important source of βAP in AD, and that microglial production of βAP may be augmented by proinflammatory stimuli or by βAP itself.

␤-Amyloid protein (␤AP) deposition is a neuropathologic hallmark of Alzheimer's disease (AD). Yet, the source of cerebral ␤AP in AD is controversial. We examined the production of ␤AP by the BV-2 immortalized microglial cell line using a sensitive enzyme immunoassay. Constitutive production of ␤AP was detected in conditioned media from unstimulated BV-2 cells. Further, production of ␤AP was induced by treatment of cultures by lipopolysaccharide (LPS) or ␤AP- (25)(26)(27)(28)(29)(30)(31)(32)(33)(34)(35) and was inhibited by the calpain protease inhibitor MDL 28170. Treatment of BV-2 cells with LPS or ␤AP-(25-35) did not affect cell-associated ␤-amyloid precursor protein levels. These findings suggest that microglia may be an important source of ␤AP in AD, and that microglial production of ␤AP may be augmented by proinflammatory stimuli or by ␤AP itself.
Alzheimer's disease (AD) 1 is a progressive neurodegenerative disorder characterized clinically by loss of cognitive function and neuropathologically by neuritic plaques. ␤-Amyloid protein (␤AP), the major extracellular protein constituent of AD plaques, consists of 39 -43 residues derived from a larger ␤-amyloid precursor protein (␤APP). The cellular source of the ␤AP deposits in AD brain is controversial (1)(2)(3). ␤APP is a ubiquitously expressed protein, present in multiple central nervous system cell types, including neurons, glia, and endothelial cells, as well as in other tissues of the body (4,5). ␤AP itself is generated in small amounts by a wide variety of cell types and is normally found in biological fluids (6 -9), but in AD there is either increased production or decreased clearance of ␤AP, or both (10). Transgenic mouse models demonstrate that neuronal overexpression of ␤APP in abnormal isoform ratios and/or of mutant ␤APP is sufficient to result in cerebral ␤AP deposits (11,12). Yet, in AD there may be multiple cellular ␤AP sources which contribute to the total cerebral ␤AP burden.
Activated microglia, which are found in and around ␤APcontaining neuritic plaques in AD brain, are candidates for ␤AP generating or processing cells (13)(14)(15)(16). In addition, the association of activated microglia with the early stages of ␤AP deposition in diffuse plaques has been demonstrated (14). Microglia are important in cerebral cytokine production, antigen presentation, and release of cytotoxic compounds and, hence, are key components in the central nervous system immune response (17). Activation of microglia in AD is part of the low-level inflammatory reaction found in the vicinity of ␤APcontaining plaques. Although microglia have been shown to express ␤APP and its encoding mRNA both in vitro and in AD brain (18 -20), secretion of ␤AP by microglia has not been demonstrated previously.
In vitro treatment of microglia with ␤AP or ␤AP fragments results in activation of the microglia and the release of inflammatory cytokines and reactive nitrogen intermediates (21,22). Recently, it has been shown that ␤AP induces its own production by smooth muscle cells (23). We treated the BV-2 immortalized murine microglial cell line (24) with the neurotoxic ␤AP-(25-35) fragment and lipopolysaccharide and measured ␤AP production using a highly sensitive enzyme-linked immunodisplacement assay (EIA). Microglial production of ␤AP in response to exogenous ␤AP would suggest an autocrine or paracrine positive feedback mechanism which could contribute to amyloidosis in AD. We also determined if BV-2 microglial cells produced ␤AP in response to the proinflammatory stimulus lipopolysaccharide, which may model some features of the inflammatory reaction in AD.
Cell Counts-BV-2 cells growing in 100-mm plates were detached by trypsinization, centrifuged, and resuspended in fresh medium. Aliquots of cells from each plate were counted three times in a hemocytometer using trypan blue exclusion, and the counts were averaged. For each treatment condition, triplicate plates were counted, and values were averaged.
Northern Analysis-Northern analysis of RNAs from treated cells was performed as described previously (25). A ␤APP hybridization probe was made by random-primed 32 P labeling of a gel-isolated BamHI fragment from the human ␤APP cDNA encompassing residues 1 to 500 of the ␤APP-751 sequence (26). This probe detected ␤APP mRNAs with and without the domain coding for the Kunitz protease inhibitor. Two DNA oligonucleotide probes complementary to mRNAs coding for mouse ␤APP-751 and mouse ␤APP-695 previously described (27) were synthesized (Operon, Alameda, CA), 5Ј terminus-labeled with 32 P using T4 polynucleotide kinase, and hybridized with Northern transfers as described previously (28). For quantitation, autoradiograms were scanned using a Pharmacia LKB ULTROSCAN XL laser densitometer, then standardized as the ratio to glyceraldehyde-3-phosphate dehydrogenase (G3PDH) mRNA hybridization signal in the same lane. The G3PDH probe was made from a gel-isolated polymerase chain reactiongenerated fragment of G3PDH cDNA (Clontech, Palo Alto, CA) and labeled as described previously (29).
Immunoprecipitation-Metabolic labeling and immunoprecipitation of ␤AP from conditioned media with the BA2 antiserum and analysis of the recovered products by 16.5% Tris/Tricine/polyacrylamide gel electrophoresis were performed as described previously (30,31). Immunoprecipitation of ␤APP from cell lysates with the BC1 antiserum raised to the cytoplasmic domain of ␤APP was also performed as described previously (30,31). Because preliminary experiments revealed that BV-2 cells showed growth inhibition after 24 h in cysteine/methioninefree medium, cells were harvested after 6 h for all metabolic labeling and immunoprecipitation experiments. Quantitation of labeled proteins was performed using a PhosphorImager (Molecular Devices, Sunnyvale, CA).
Western Analysis-To examine cell-associated ␤APP after 24 h of treatment with LPS, Western analyses were performed on triplicate BV-2 cultures. Cells were lysed using RIPA buffer (50 mM Tris-HCl, pH 7.5, 150 mM NaCl, 1% deoxycholate, 0.1% SDS, 1% Triton X-100), and the lysate was clarified by centrifugation at 39,000 ϫ g for 15 min. Total protein was quantified using the Lowry method, and equal amounts of protein were loaded into each lane of an 8% polyacrylamide gel, followed by transfer to an Immobilon-P membrane (Millipore, Bedford, MA). Immunodetection was performed using the BC1 antiserum and the peroxidase-antiperoxidase method. Briefly, membranes were blocked with normal swine serum and incubated overnight at room temperature in BC1 at a 1:100 dilution in 50 mM Tris-buffered saline (TBS), pH 7.6, containing 1% nonfat dry milk. Membranes were then incubated in swine anti-rabbit serum (1:50) for 30 min, followed by a rabbit peroxidase/antiperoxidase (1:400) for 30 min. Finally, bound antibodies were visualized with diaminobenzidine, 0.5 mg/ml in 50 mM Tris, pH 7.6, with 15 l of H 2 O 2 for 7 min. Membranes were washed 3 times in TBS for 5 min after each antibody incubation.
␤AP Enzyme Immunoassay-␤AP was detected in conditioned media by an antigen displacement enzyme immunoassay (EIA). Flat-bottom 96-well microtiter plates (Corning, Cambridge, MA) were precoated with goat anti-rabbit IgG (Zymed, South San Francisco, CA) by incubating each well for 2 h at 37°C with 0.5 g of antibody in 0.2 ml of 0.1 M sodium bicarbonate buffer, pH 8.5. The plates were washed four times in TNT buffer (0.05 M Tris, pH 7.5, 0.15 M NaCl, 0.05% Tween 20) and blocked for 1 h at 37°C in TNTB buffer (TNT buffer containing 1% BSA). Aliquots of 0.1 ml of antiserum BA1 (1:5000 dilution in TNTB buffer) were added to each well, followed by the immediate addition of 0.1 ml of synthetic ␤AP-(1-40) standard (Bachem) or the unknown. The dilutions of the standard and the unknown were prepared in sample buffer (0.1% BSA, 0.1% Triton X-100 in water). The antibody-antigen mixture was co-incubated overnight at 4°C. Aliquots of 0.05 ml containing 1 ng of biotinylated ␤AP-(1-40) were added to each well, and the mixture was incubated for an additional 1 h at 4°C. The contents of the plate were discarded, and 0.2-ml aliquots of horseradish peroxidaselabeled streptavidin (Zymed) at 1:3000 dilution were added and incubated for 1 h at 4°C. The plates were washed five times in TNT buffer, and the bound enzyme activity was detected by the addition of 0.2-ml aliquots of substrate (3,3Ј,5,5Ј-tetramethylbenzidine liquid substrate system, Sigma). The enzyme reaction was stopped by the addition of 0.1 ml of 0.5 M sulfuric acid, and the absorbance was measured at 450 nM in a microplate reader. Data were expressed as nanograms of ␤AP per plate (10 ml of conditioned medium). Biotinylated ␤AP-(1-40) was prepared as follows: 0.5 mg of synthetic ␤AP-(1-40) was reacted with a 10-fold excess (mole to mole) of aminohexanoyl-biotin-N-hydroxysuccinimide ester (Zymed) in sodium bicarbonate buffer, pH 8.5, for 3 h at room temperature. Tween 20 was added to the reaction mixture to make a final concentration of 0.05%, and the reaction mixture was dialyzed overnight against 0.05 M Tris buffer containing 0.05% Tween 20. The dialysis step removed the unreacted biotin and concentrated the biotinylated peptide. The biotinylated peptide was stored in the presence of 1% BSA at Ϫ70°C. Conditioned media (3 ml) from BV-2 microglial cells were passed through Sep-Pak C18 cartridges and the 50% CH 3 CN eluate collected and dried as described previously (10). The 50% CH 3 CN eluate was resuspended in sample buffer. The EIA does not detect the ␤AP- (25)(26)(27)(28)(29)(30)(31)(32)(33)(34)(35), but rather full-length ␤AP species. The epitope recognized by the BA1 antiserum was mapped to the carboxyl terminus of ␤AP-(1-40), the principal ␤AP species secreted by cultured cells (32). The specificity of the BA1 antiserum was further verified by reactivity with a collection of ␤AP synthetic peptides. The EIA also detected ␤AP-(1-42), but with 10-fold less efficiency than for ␤AP- .
Statistical Analysis-A one-way analysis of variance followed by pairwise comparisons among the various treatment means using Fisher's Least Significant Difference test was used for comparison among three or more means. Analysis using the nonparametric Kruskal-Wallis H test followed by pairwise Mann-Whitney U tests gave identical results (results not shown). For comparison of two means, unpaired t tests were used. All p values presented are for two-tailed tests.

RESULTS
␤AP Secretion-Analysis of conditioned media using the EIA demonstrated that 24-h treatment with either aggregated ␤AP- (25)(26)(27)(28)(29)(30)(31)(32)(33)(34)(35) or LPS alone or in combination increased ␤AP production by BV-2 cells. A one-way analysis of variance showed an overall difference in mean ␤AP levels in BV-2 conditioned media among the four treatment conditions (control, LPS, ␤AP, and LPS plus ␤AP; F ϭ 17.67; df ϭ 3,8; p Ͻ 0.001; Fig. 1A). ␤AP levels in culture media from LPS-stimulated cells were 1.85 times greater than those in media from control cells, a statistically significant difference. ␤AP levels in media from cells stimulated with ␤AP or LPS plus ␤AP were also significantly increased over control levels. The LPS plus ␤AP condition did not differ significantly from either ␤AP alone or LPS alone.
The LPS-induced increase in ␤AP in conditioned media was significantly attenuated by administration of the calpain protease inhibitor MDL 28170 (experiment 3; Fig. 1B). The MDL 28170 inhibitor has been reported previously to selectively inhibit ␤AP formation by blocking the action of ␥-secretase activity (31). A significant overall difference in mean ␤AP levels in media among the four treatment conditions was detected using a one-way analysis of variance (control, LPS, MDL 28170, and LPS plus MDL 28170; F ϭ 34.9, df ϭ 3,11; p Ͻ 0.0001). Pairwise comparisons showed that both MDL 28170 alone and LPS plus MDL 28170 resulted in significantly less ␤AP in conditioned media compared to control or LPS values. There was no significant difference between the MDL 28170 and the LPS plus MDL 28170 treatments. These data further validate that ␤AP is being detected and that ␤AP production is stimulated by LPS treatment.
Treatment of BV-2 cells with MDL 28170 at a concentration of 35 M gave no indication of toxicity, either by morphological inspection or with the trypan blue viability assay. In addition, yields of total RNA were comparable between culture conditions, and Northern analysis indicated no change in G3PDH mRNA in cultures treated with MDL 28170, a further indication that no toxic effect had occurred (data not shown).
Control assays were performed on blank media which had not been exposed to cultures, or blank media with LPS or MDL 28170. These samples showed no evidence of ␤AP immunoreactivity. When conditioned media were collected from control cells after 24 h and aggregated ␤AP-(25-35) was added, there was no difference from other control cultures in ␤AP detected by the EIA. LPS, MDL 28170, and ␤AP-(25-35) added to conditioned media did not affect the performance of the EIA (data not shown).
␤AP Immunoprecipitation-Immunoprecipitation of condi-tioned medium from BV-2 cells with the BA2 antiserum to ␤AP followed by Tris-Tricine gel electrophoresis revealed the presence of the characteristic ␤AP signal at approximately 4 kDa (Fig. 2). In medium harvested after 6 h, LPS treatment resulted in an approximately 2-fold increase in ␤AP signal, providing visual verification of the results obtained using a different ␤AP antiserum (BA1) and a different method of analysis (EIA).
Morphologic Effects-Treatment of BV-2 cells with aggregated ␤AP-(25-35) resulted in visible precipitates under phase contrast microscopy in the cultures, indicating that aggregation of the peptide had occurred. Following incubation with the aggregated peptide, the BV-2 cells began to clump within 24 h (Fig. 3C). LPS alone did not cause clumping (Fig. 3B), but appeared to enhance the ␤AP effect (Fig. 3D). MDL 28170 at a concentration of 35 M had no morphological effect (data not shown).
Cell Counts-The effects of LPS treatment were assessed with respect to cell proliferation. This was done so as to exclude increased cell numbers as being responsible for increased ␤AP in BV-2 conditioned media. No difference was found. After 6 h of treatment with LPS, the mean cell count from three plates was 5.2 ϫ 10 5 cells/ml (S.E. ϭ 0.7), whereas the mean count in control cultures was 4.8 ϫ 10 5 cells/ml (S.E. ϭ 0.3). After 24 h of LPS treatment, the mean cell count from three plates was 6.7 ϫ 10 5 cells/ml (S.E. ϭ 0.1), whereas the mean count from control plates was 8.4 ϫ 10 5 cells/ml (S.E. ϭ 0.2). Control cultures grown in cysteine/methionine-free medium for 6 h showed a mean cell count of 5.0 ϫ 10 5 cells/ml (S.E. ϭ 0.8).
␤APP Immunoprecipitation and Western Analysis-Immunoprecipitation of ␤APP from BV-2 cell lysates and Western analyses of total cellular protein revealed ␤APP species with molecular masses in the range of 95-130 kDa depending on the cellular metabolic state, isoform(s), and the extent of ␤APP glycosylation. After 6 h of metabolic labeling, immunoprecipitation showed unglycosylated ␤APP with a molecular mass of approximately 98 kDa, as well as partially glycosylated forms with a molecular mass of approximately 105-115 kDa (Fig. 5). It is likely this pattern reflects labeling of primarily immature  (25)(26)(27)(28)(29)(30)(31)(32)(33)(34)(35), and LPS plus ␤AP-(25-35) were significantly greater than the control mean (p Ͻ 0.05 for all three comparisons). There were no significant differences among the means for ␤AP- (25)(26)(27)(28)(29)(30)(31)(32)(33)(34)(35), LPS, and ␤AP-(25-35) plus LPS. Data are expressed as nanograms of ␤AP per plate (10 ml of conditioned medium). B, EIA results from experiment 3, three replicates for MDL 28170, four replicates for control, LPS, and LPS plus MDL 28170, means with S.E. The mean for LPS was significantly greater than the control mean, whereas the mean for MDL 28170 was significantly less than the control mean (p Ͻ 0.05 for both). The mean for LPS plus MDL 28170 was significantly less than the control mean and the mean for LPS alone (p Ͻ 0.05 for both). The MDL 28170 and LPS plus MDL 28170 treatment means were not significantly different. Data are expressed as nanograms of ␤AP per plate (10 ml of conditioned medium). forms of ␤APP due to the relatively short 6-h interval and the nutritionally deficient medium. In contrast, ␤APP species detected by Western analyses in total cellular protein harvested after 24 h were largely 105-115-kDa partially glycosylated forms, as well as mature 130-kDa ␤APP (Fig. 6). This pattern is similar to that observed by Monning et al. (33) who immunoprecipitated unlabeled ␤APP from BV-2 cells.
Immunoprecipitation of ␤APP from cell lysates after metabolic labeling showed no difference in ␤APP levels between cells treated with LPS for 6 h and control cells (Fig. 5). Likewise, triplicate Western analysis of cells treated with LPS for 24 h showed no difference in ␤APP levels in comparison with controls (Fig. 6). Therefore, increased ␤AP in conditioned medium was not simply a result of higher levels of cellular ␤APP.
Treatment of BV-2 cells with LPS also increased ␤AP in conditioned media. LPS treatment did not result in an increase in cell numbers, so the increased ␤AP in conditioned media must be attributed to increased production of ␤AP per cell. The increased ␤AP induced by LPS was observed after both 6 and 24 h of treatment. Combined treatment with ␤AP and LPS also increased ␤AP in conditioned media compared with controls, although the increase was not significantly greater than that observed with ␤AP or LPS alone.
Although there was an increase in ␤APP mRNA after 24 h of LPS treatment, no significant change in ␤APP mRNA was observed after 6 h of treatment, despite the fact that ␤AP was increased at the 6-h time point. Further, immunoprecipitation and Western analyses for ␤APP showed no increase in ␤APP after either 6 or 24 h of LPS treatment. Thus, the LPS-induced increase in ␤AP production does not appear to be driven by increased ␤APP expression. A previous study suggested that when BV-2 cells are cultured on dishes coated with laminin or fibronectin, LPS treatment results in an increase in ␤APP expression (33). However, in agreement with our results, no significant change in ␤APP expression was observed in that study after LPS treatment of BV-2 cells cultured in uncoated plastic dishes.
We used a number of different methods to confirm the modulation of ␤AP production in BV-2 cells. A ␤AP-(1-40)-selective EIA was employed most frequently in this study. In addition, using immunoprecipitation, we detected a protein that migrated at the ␤AP position. The authenticity of the ␤AP was further demonstrated by treatment of BV-2 cells with the calpain protease inhibitor MDL 28170. This agent, which has previously been shown to inhibit the formation of ␤AP (31), resulted in a significant decrease in immunoreactivity detected in BV-2 conditioned medium.
These findings support the hypothesis that inflammatory activation of microglia in AD leads to increased microglial production of ␤AP which is deposited in plaques. Although previous studies have indicated that microglia express ␤APP (18,20,33), heretofore it has not been known if immune activation of microglia cells results in increased ␤AP production. Further, we present evidence that exogenous ␤AP may stimulate microglial ␤AP production. This suggests that in AD increased extracellular ␤AP may induce microglia to produce additional ␤AP, which in turn could lead to further ␤AP production by microglia or other cells via an autocrine or paracrine positive feedback mechanism. Because ␤AP has been shown to increase microglial inflammatory cytokine secretion and production of reactive nitrogen intermediates (21,22), ␤AP produced by microglia may also augment the local inflammatory response surrounding the neuritic plaque, which may lead to neuronal dysfunction or injury.
It is significant that immunologic stimulation of BV-2 cells with LPS resulted in a large increase in ␤AP in conditioned media. BV-2 cells have many of the features of reactive, rather than resting microglia cells (24,38). Reactive macrophages and microglia are known to contain and to secrete very high levels of proteolytic enzymes (39). The secretases which cleave ␤APP to amyloidogenic fragments may be among the enzymes which are present at high levels in reactive microglia, and their activity may be enhanced after immunologic stimulation. The initiating event for ␤AP deposition in AD remains obscure as does the stimulus resulting in signs of immunologic activation in neuritic plaques. However, the significant increase in ␤AP produced by BV-2 cells when exposed to an immunogenic stimulus such as LPS suggests that activated microglia may be a major source once the pathological immune process is initiated in AD.
In conclusion, these findings indicate that microglial cells activated by inflammatory stimuli may play an important role in ␤-amyloidosis in AD. Anti-inflammatory agents which delay the onset of Alzheimer's pathology (40, 41) may work by inhib-iting microglial ␤AP production and the secretion of microglial cytokines and other proinflammatory products. In addition, we demonstrate that ␤AP itself may act as an exogenous stimulus to its own production by cultured microglial cells. If this occurs in AD brain, secreted ␤AP may have important autocrine or paracrine effects which augment its own production. Further understanding of the role of microglia and cerebral inflammatory processes in AD may lead to a clearer understanding of the pathogenesis of this devastating disorder. In addition, development of drugs targeted to microglial proteases may provide a new line of treatment.