Journal of Neuroscience Research 43:19&202 (1996)

Scavenging of Alzheimer’s Amyloid P-Protein by Microglia in Culture M.D. Ard, G.M. Cole, J. Wei, A.P. Mehrle, and J.D. Fratkin Departments of Anatomy (M.D.A., J.W., A.P.M) and Pathology (J.D.F), University of Mississippi Medical Center, Jackson; Department of Medicine, University of California at Los Angeles and Geriatric Research, Education, and Clinical Center, V.A. Medical Center, Sepulveda (G.M.C), California

Deposits of amyloid P-protein (AP) form the cores of the pathological plaques which characterize Alzhei- mer’s disease. The mechanism of formation of the deposits is unknown; one possibility is failure of a clearance mechanism that would normally remove the protein from brain parenchyma. This study has investigated the capacity of the central nervous sys- tem (CNS) phagocytes, microglia cells, to clear exog- enous AP,,, from their environment. Cultured mi- croglia from adult rat CNS have a high capacity to remove A P from serum-free medium, shown by im- munoblotting experiments. A P from incubation me- dium was attached to the cell surface and could be identified by immunocytochemistry at the light or electron microscopic (EM) level; by EM, A P also ap- peared in phagosome-like intracellular vesicles. Light microscopic immunocytochemistry combined with computer-assisted image analysis showed that cells accumulated A P within 24 hr. from culture medium containing from 1 to 20 pg/ml AP.

Microglial accumulation of A P was substantially reduced in the presence of fetal bovine serum. Addi- tion of the protease inhibitor leupeptin to incubation medium with serum resulted in accumulation of AP in a membrane-bound intracellular compartment, but not at the cell surface. The increase in intracel- lular accumulation in the presence of the protease inhibitor indicates a microglial capacity for intracel- lular degradation of A P in the absence of inhibition. The change from predominantly cell-surface accumu- lation in serum-free medium to predominantly intra- cellular accumulation with serum may be explained by the presence in serum of carrier proteins that com- plex with A P and target it to cell surface receptors capable of stimulating endocytosis.

Microglia were also cultured on unfixed cryostat sections of human brain tissue containing Alzheimer’s plaques. Very little AP from the tissue was accumu- lated by the cells, although cultured microglia were found in direct contact with anti-AP immunopositive plaques. Possibly AP in tissue sections was complexed

0 1996 Wiley-Liss, Inc.

with other proteins which either inhibited its uptake by microglia or enhanced its proteolysis, preventing cellular accumulation of immunostainable AP.

The results indicate that cultured microglia effec- tively remove A P from tissue culture medium and from the surface of the dish and concentrate mono- mer and aggregates of A P either on the cell surface or intracellularly. This process may be modified by pro- teins present in Alzheimer’s brain sections. 0 1996 Wiley-Liss, Inc.

Key words: amyloid P-protein, APIa2, central ner- vous system phagocytes, microglia

INTRODUCTION

Since genetic mutations in P-amyloid protein (AP) or its precursor (APP), as well as excess gene dosage in trisomy 21, are strongly linked to Alzheimer’s disease, deposition of AP in plaques may be not only a patho- logical hallmark resulting from Alzheimer’s but also a cause of the disease. Therefore, understanding the cel- lular and molecular mechanisms which produce and maintain P-amyloid plaques could facilitate progress in controlling the disease. AP is a normal, secreted product of cells and is found in CSF (Haass et al., 1992; Seubert et al., 1992; Shoji et al., 1992; Busciglio et al., 1993). To date, increases in AP production or length have been implicated in accelerating AP deposition. However, in principle, defects in A@ clearance may be equally likely to cause accelerated AP deposition.

Microglia cells are consistently associated with the compact and neuritic plaques found in pathological spec- imens of Alzheimer’s disease brains (Terry et al., 1964; Itagaki et al . , 1989; Vaughan and Peters, 1981, in rat

Received November 30, 1994; revised and accepted June 30, 1995.

Address reprint requests to March D. Ard, Department of Anatomy, University of Mississippi Medical Center, 2500 N. State St., Jackson, MS 39216.

Scavenging of AP by Microglia 191

transferrin, 50 pg/ml ascorbate, and penicillin-strepto- mycin. Basic fibroblast growth factor (FGF) was added at 0.5 ng/ml. Enrichment of the cultures for microglia was achieved by differential adhesion: the cell suspen- sion was plated onto polylysine-coated glass coverslips (12 coverslips for 1 spinal cord) and allowed to attach overnight. Few oligodendrocytes adhered, and the next day the coverslips were rinsed to remove floating cells. Cultures were 94-99% microglia: 94% of cells plated in C medium and 99% of cells plated in GN medium were positive for OX-42 monoclonal antibody to complement receptor C3b (Serotec, distributed by Harlan Bioproducts for Science, Inc., Indianapolis, IN). Cells plated in C medium were also 87% positive for Bandeirueu simplic- ifolia isolectin B4 (Sigma Chemicals); and cells in GN were also tested with DiI-conjugated acetylated low den- sity lipoprotein, a label for scavenger receptor, and found to be 93% positive.

Materials Amyloid P-protein was a synthetic, human se-

quence, 42 amino acid peptide prepared and purified in 70% formic acid on P4 by Dr. N. Ling (Whittier Insti- tute, La Jolla, CA). Lyophilized AP was dissolved in dimethyl sulfoxide (10 pg/pl) and frozen in aliquots at -80°C. Immediately before addition to cultures, each aliquot was thawed and diluted 1:l with 35% acetoni- trile/O. 1% trifluoroacetic acid; this solution was added to culture medium to a final concentration of 1, 5, or 20 pg/ml. AP was detected immunocytochemically using polyclonal antiserum raised in rabbits to human sequence

Human brain tissue from hippocampus and supe- rior temporal cortex of two Alzheimer’s cases and one non-Alzheimer’s control was supplied by the National Neurological Research Specimen Bank (VAMC Wads- worth Division, Los Angeles, CA 90073, sponsored by NINDS/NIMH, National Multiple Sclerosis Society, He- reditary Disease Foundation, Comprehensive Epilepsy Program, Tourette Syndrome Association, Dystonia Medical Research Foundation, and Veterans Health Ser- vices and Research Administration, Department of Vet- erans Affairs). Cryostat sections of unfixed, frozen tissue were thaw-mounted onto polylysine-coated coverslips, and microglial cell suspensions were plated onto the sec- tions within 2 hr after thaw-mounting. Cultures on sec- tions were incubated for 2-5 days.

Methods Immunostuining was carried out by standard meth-

ods, antibody preparations diluted 1: 100, incubations 1 hr at room temperature. Cell surface antibodies (OX-42) were applied to living cultures before fixation. Cultures were fixed after staining in 4% paraformaldehyde in phos-

4 3 - 4 ” or AP14-*4.

brain). Electron micrographs from such specimens show an association between microglia and P-amyloid fibrils, indicating that microglia may phagocytose P-amyloid (Itagaki et al., 1989; McGeer et al., 1992); on the other hand, electron micrographs have been interpreted as meaning that microglia may secrete amyloid P-protein (Terry et al., 1964; Wisniewski et al., 1989; Frackowiak et al., 1992). Ingestion of AP has been shown immuno- histochemically following its injection into rat hippo- campus, and phagocytes loaded with immunostained ma- terial apparently migrated to blood vessel walls and to the ventricular surface, as if engaged in clearing the ex- ogenous A@ from the brain parenchyma (Frautschy et al., 1992).

These cumulative observations of microglia in situ suggested the present cell culture study to determine the capability of microglia to clear AP from their environ- ment. Synthetic human sequence AP,_,, was added to culture medium at relatively low concentrations, and cel- lular accumulation of immunostainable material occurred (Ard et al., 1993). In contrast, in a series of experiments in which microglia were cultured on cryostat sections of Alzheimer’s disease brain tissue, the cells were observed to accumulate little or no AP from plaques in the tissue. Microglial clearance of soluble AP and of the earliest deposited aggregates may be an important defense against AP plaque accumulation.

MATERIALS AND METHODS Cell Culture

Adult microglia from rat spinal cord were cultured by the method developed by Wood and Bunge (1 986) for oligodendrocytes. The whole spinal cord, beginning ros- trally at the cervical enlargement, was removed from a 2-3-month-old female rat after euthanasia. The me- ninges were removed and the spinal cord sliced, incu- bated in trypsin, mechanically dissociated, and passed through 15 p m pore size nylon mesh to remove undis- sociated clumps. Cells were then separated from myelin debris on a 25% Percoll (Sigma Chemicals, St. Louis, MO) gradient formed at 30,000g for 45 min. The Percoll layer containing both oligodendrocytes and microglia was situated above a thin band of red blood cells and below a thick band of myelin debris. The cells were washed to remove Percoll, passed again through a nylon mesh, and plated in either C medium (Eagle’s minimal essential medium plus 10% heat-inactivated fetal bovine serum plus penicillin-streptomycin), or serum-free GN medium, based on media formulations for glia and neu- rons by Bottenstein (1984) and Michler-Stuke et al. (1984). GN consisted of 1:l DMEM:Ham’s F-12 plus 5 nM hydrocortisone, 5 pg/ml insulin, 100 pM putrescine, 20 nM progesterone, 30 nM sodium selenite, 50 pg/ml

Fig. 1. A@ is accumulated by microglia from incubation me- print-out of targets selected from the negative of D €or mea- dium. A, C, and E are a single field of a microglia culture surement of anti-AP immunostaining of cells. Deposits of im- incubated with 1 pg/ml AP for 24 hr. The culture shown in B, munopositive material on the substratum outside cells were not D, and F was incubated with 5 pg/ml AP for 24 hr. A and B , measured. Cells are shown as outlines; targets measured are phase contrast; C and D, FITC immunofluorescence with anti- solids. Comparison of C to D illustrates increased AP accu- A@; E, TRITC immunofluorescence with OX-42 anti-comple- mulation with increased concentration of exogenous A@. ment receptor to identify the cells as microglia; F, computer X 300.

Scavenging of AP by Microglia 193

Average Area of lmmunostaining Per Cell, In pm2 phate buffered saline. Intracellular antigens were immu- nostained after fixation of cultures in the same fixative. Cell membranes were permeabilized in cold acid-alcohol (5% acetic acid, 95% ethanol at 0°C). Secondary anti- bodies were fluorescein isothiocyanate-conjugated or tetra-rhodamine isothiocyanate-conjugated, affinity puri- fied goat anti-rabbit or goat anti-mouse IgG, from Jack- son Immunoresearch (West Grove, PA) or Cappel (Durham, NC).

Computer-assisted image analysis of immuno- stained cultures provided quantitation of AP accumula- tion by microglia. A predetermined sample area of each culture was photographed. From the negatives, the total area of immunostaining that was associated with cells was measured (MCID image analysis software by Imag- ing Research, Inc., St. Catherine’s, Ontario), and the total number of cells in the sample (about 100) was counted. Targets for measurement were selected by the human observer so that only cell-associated immunopo- sitive deposits, not deposits on the substratum outside cells, were counted. The final result was expressed as average area of immunostained material per cell.

Electron microscopy was performed on cultures fixed in 2% glutaraldehyde, postfixed in 2% OsO,, stained en bloc with uranyl acetate, and embedded in Polybed (Polysciences, Inc., Warrington, PA) or Durcu- pan (Electron Microscopy Sciences, Fort Washington, PA). Thin sections were immunostained after etching in hydrogen peroxide, using the same polyclonal anti-AP antibodies as for light microscopy and 15 nm gold-la- beled goat-anti-rabbit secondary antibody (Amersham, Arlington Heights, IL). Grids were counterstained with lead citrate.

Immunoblotting was employed to analyze clear- ance of AP from culture medium. Conditioned medium was collected with a protease inhibitor cocktail consist- ing of 0.6 pg/ml pepstatin, 5 pg/ml leupeptin, 1 mM phenylmethylsulfonyl fluoride, and 2 mM EDTA. Cells were rinsed once and scraped into lysis buffer (0.85% NaCl plus 0.05 M Tris, pH 8.0, with protease inhibitor cocktail as above) and sonicated. Aliquots were taken for total protein analysis. Lysates were adjusted to 1 pg/ml. Fifty microliters of each sample, either conditioned me- dium or cell lysate, was mixed with Laemmli sample buffer, electrophoresed on lO/l6% Tris Tricine gels (Schagger and von Jagow, 1987), and blotted to Immo- bilon membranes (Millipore, Bedford, MA). The blots were blocked, then developed with a monoclonal anti- body to free P,p40 (10G4 antibody, which recognizes P5p,3) using an enhanced chemiluminescence kit (ECL kit, Amersham) according to manufacturer’s instruc- tions. Fluorograms of immunoblots were densitometri- cally scanned on an image analysis system (Imaging Re- search, Inc.).

160

140 0

8

0

0

2o 0 0.0 3 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0 4.5 5.0 5.5 6.0

.ug/ml of ApwzAdded To Culture Medium

Fig. 2. Image analysis results of immunostaining show AP accumulation by microglia in serum-free medium.

RESULTS Microglial Clearance of AP From Serum-Free Medium

Adult rat microglia, cultured overnight, were pre- sented with soluble API-42 for a period of 24 hr at 37°C. Cells were then fixed and immunostained with antibodies to AP and a fluorescent secondary antibody. The amount of immunostained material accumulated by the cells was quantitated by image analysis of photographic negatives. Accumulation of immunostained material occurred over a range of 1 pg/ml to 5 pg/ml AP (Figs. 1,2). At lower concentrations, i.e., 0.1 and 0.2 pg/ml, the amount of positive immunostaining per cell was not above the back- ground level seen in cells not exposed to exogenous AP (Fig. 2). In contrast, at 5 pg/ml, an average of 70% of cells were positive for anti-AP immunostaining, and in many cells the area of immunostained material was al- most as large as the cell itself (Fig. 1D). Higher concen- trations of AP were therefore not tested. Immunostained material appeared not only in the cells but also attached to the polylysine substratum. Aggregates of protein on the substratum apparently could be phagocytosed by mi- croglia, since clear areas were often seen surrounding individual cells. Aggregates were also found in immu- noblotted cell lysates (Fig. 4).

Although microglia, like many other cells, have been shown to synthesize amyloid precursor protein (Haass et al., 1991; Banati et al., 1993; but see also Scott et al., 1993), under the conditions of this study there was little or no accumulation of immunostainable material in microglia cultured without exogenous AP protein. Pre- sumably, low levels of AP are effectively cleared or require long time periods to accumulate to detectable levels.

Most of the accumulated AP appeared to be on the surface of the microglia, since immunostaining results

194 Ard et al.

Fig. 3. Material immunolabeled with anti-AP and 15 nm gold particles (arrows) was distributed predominantly at the cell surface of microglia incubated with 5 pg/ml AP in serum-free medium. Although AP aggregates mostly were not endocy- tosed, the cells were actively phagocytic and contained a va- riety of inclusions that were not immunolabeled. A small

amount of immunolabeled material was also found in intracel- Mar granules (C, from right half of cell shown in A). The immunogold labeled material was not fibrillar in structure (B, from left half of cell shown in A). Magnification of A = X 13,000; B and C = X 22,500.

with living cell cultures were similar to those obtained with fixed and permeabilized cells.

Electron microscopy confirmed that the immuno- stained material was located predominantly at the cell surface, with only a minor proportion in vesicles inside the cells (Fig. 3). Gold-conjugated secondary antibody following anti-AP primary labeled aggregated material of medium electron density, without any apparent fibril- lar structure.

In one experiment to test for fibrillar conformation of the AP-like material accumulated by the microglia,

the cells were stained with Congo red. The results were negative (data not shown), in agreement with the images seen in electron micrographs. As a positive control, Alz- heimer’s brain tissue sections were stained with Congo red at the same time and gave positive results (apple- green birefringence of plaque amyloid).

Immunoblotting of conditioned media and cell lay- ers showed that AP was cleared from the culture medium and acquired by the microglial cells (a representative blot is shown in Fig. 4). Serum-free medium containing 5 p,g/ml or 10 pg/ml AP showed a marked reduction in

Scavenging of AP by Microglia 195

Fig. 4. Immunoblotting of a microglia culture incubated with 10 kg/ml AP shows AP clearance from the medium and ac- cumulation by the cells. Samples of conditioned media and cells were electrophoresed on a 10116% step gradient Tris- tricine gel, transferred to PVDF membrane, and probed with monoclonal antibody to human AP,-,<). Lane 1, fresh medium; lane 2, fresh medium with AP; lanes 3,4, conditioned medium (CM) from cells incubated without A@; lanes 5,6, CM from cells incubated with AP; lanes 7,8, medium with AP incubated without cells; lane 9, cell lysate, cells incubated without AP; lane 10, cell lysate, cells incubated with AP; lane 11, lysis buffer scraped from dish incubated with medium with AP, without cells. Arrowhead indicates AP and arrow indicates low molecular weight aggregates. Higher molecular weight aggre- gates appear in cell lysate of cells incubated with AP (lane 10).

both monomer and aggregate AP bands after incubation with cells for 48 hours (Fig. 4, lanes 5,6, with cells, compared to lanes 7,8, medium incubated without cells), while monomer and higher molecular weight aggregates became concentrated in the cell layer (lane 10). Medium containing AP was incubated either with cultured micro-

Fig. 5 . The histogram depicts results of densitometry of the immunoblot shown in Figure 4.

glia, o r in a polylysine-coated culture dish without cells, so that incubation conditions were equivalent for cell- conditioned and for control medium. AP adsorbed to the substratum of the control dish was scraped off and ana- lyzed (lane l l ) for comparison to the cell layer (lane lo), since the cell layer included both cells and material bound to the substratum. Only aggregates up to 46 kDa are included in Figure 4; some higher molecular weight aggregates were also present, particularly in lanes 7,8, and 10; but nonspecific luminescence on the fluorograms precluded accurate analysis of these regions.

Aggregated AP was especially prevalent in the cell layer. Densitometric scanning of the immunoblot shown in Figure 4 provided a semi-quantitative comparison of AP in 4 kDa bands with that in higher molecular weight aggregates of 18 kDa and 30 kDa (Table I). The densi- tometric ratios of aggregates to monomer in media and cell lysates are illustrated by the histogram, Figure 5 .

TABLE I. Densitometric Scanning of Immunohlot Shown in Figure 4 (Density X Area [arbitrary units] for Bands at Given Molecular Weights)

Description Lane 4 kDa 18 kDa 30 kDa Total*

Cell-conditioned medium 5 12.2 6.8 1.5 Fresh medium with AP 2 35.7 10.7 2.0

With AD 6 16.3 5.9 0

With AP 8 37.0 31.3 0 (Average) (39.7) (37.8) (0.3) 78

Cell lysate with substratum-bound AP 10 57.6 134.6 15.1 207 Substratum-bound AB, without cells I I 27.1 46.8 0.6 74

(Average) (14.3) (6.6) (0.7) 22 Medium incubated without cells 7 42.3 38.3 0.6

*Statistical comparison of the numbers in this column, total densities of anti-AD-positive bands, using the x2 distribution shows significant clearance of AB from the medium and accumulation by cells when medium containing AD is incubated with microglia. P < 0.001.

196 Ard et al.

TABLE 11. x2 Contingency Table*

Medium Cell lysate Totals

Without cells 78 74 152 With cells 22 207 229

*The distribution of AD incubated with cells differs from that without cells. P < 0.001.

The ratio of aggregate to monomer in the cell layer was 5.2 times higher than that in cell-conditioned medium, and 2.6 times higher than that in medium incubated with- out cells, indicating accumulation of aggregates by cells. Aggregates may form in the medium, or aggregation may occur within the cell layer.

The x2 statistic was used to compare the proportion of AP remaining in the medium to that attached to cells and substratum in each incubation condition (Table 11). The x2 statistic is useful because the comparison of dis- tributions of AP does not depend on the total amounts of immunostaining product being the same in the two con- ditions. In medium incubated without cells, about half (49%) of the AP attached to the polylysine substratum during the 48 hr incubation. When microglia cells were present, 90% of the AP attached to the cells and substra- tum. This is a significant difference (P < 0.001) in dis- tribution, showing clearance of AP from the medium by microglia. Similar results were seen with 5 p,g/ml AP added to the incubation medium. Clearance of AP at these concentrations suggests high capacity uptake by microglia.

Microglial Clearance of AP in the Presence of Serum

Culturing microglia in medium containing 10% fe- tal bovine serum greatly altered the results of exposure to AP in culture medium. In seven separate experiments, AP was added at concentrations of 1, 2, 5 , or 20 pg/ml; accumulation of immunostained material by microglia was near minimal in four experiments, but there was considerable accumulation in one experiment at 5 pg/ml, and some accumulation of immunostainable material in two others (Fig. 6). At high concentrations, the tendency of AP to aggregate may significantly affect results. In- cubations with AP were for 48 hr. A single batch of heat-inactivated serum was used throughout the series.

In medium with serum, addition of the protease inhibitor leupeptin (100 pg/ml; Wolozin et al., 1992) allowed microglia to accumulate AP intracellularly (Fig. 7). Eight experiments were done, varying the concentra- tion of AP from 1 to 20 pg/ml. At each concentration of AP, accumulation of immunostained material was ob- served in the presence of leupeptin (Fig. 6); a single experiment, at 20 pg/ml, was an exception in that the

Average Area of lmmunostaining Per Cell, In p2

140 0 Serum With Leupeptin 120 1 0 Medium With Serum

100 4 E, 80 i f 60 1 m 0

0

-~ 0 2 4 6 8 10 12 14 16 18 20 22 24

pglml of Ap,,,Added To Culture Medium

Fig. 6. Image analysis results of immunostaining were some- what variable when microglia were incubated with A@ in the presence of serum (squares). Generally, little A@ was accumu- lated by the cells, with one exception (cells incubated with 5 pg/ml AP, squares). Addition of the protease inhibitor leupep- tin to the medium (diamonds) resulted in marked accumulation of A@, again with one exception (cells incubated with 20 kg/ml AP, diamond). With leupeptin, the average area of im- munostaining per cell was less than in serum-free medium.

amount of material accumulated was much lower than expected.

Only AP 1-42 accumulated intracellularly when added with leupeptin; microglia accumulated little AP1-40 even when leupeptin was present (Fig. 7E,F).

Electron microscopy of cultures incubated with AP in the presence of serum and leupeptin showed the im- munostained material to be located intracellularly , in variably shaped, electron-dense, membrane-bound endo- somes (Fig. 8). The labeled material did not have an obvious fibrillar structure.

Microglia Cultured on Tissue Sections Since purified AP,-,, may not be equivalent to the

form in which P-amyloid protein is presented to scaven- ger cells in vivo, the ability of cultured microglia to phagocytose amyloid from Alzheimer’s brain tissue was also investigated. Unfixed cryostat sections of temporal cortex or hippocampus from confirmed cases of Alzhei- mer’s disease were thaw-mounted onto polylysine- coated coverslips, and freshly prepared microglia cell suspensions were plated onto the tissue sections within 2 hr. This experiment was repeated 4 times using culture medium with serum, with and without the addition of leupeptin or ammonium chloride to inhibit lysosomal ac- tivity. The fifth and sixth repetitions used serum-free medium. In some experiments some sections were pre- treated with 50% formic acid for 5 min to increase ac- cessibility of P-amyloid in the tissue. Microglia were made visible on the tissue sections by phagocytosis of 3

Fig. 7. Microglia cultured with 10% serum accumulated little AP unless leupeptin was added to the incubation medium. A, C, and E are phase contrast images of microglia incubated with serum and 5 pg/ml AP; B, D, and F are fluorescent images of the same cells immunostained with anti-A@. In A and B, few cells accumulated immunostainable AP, whereas with the ad-

dition of leupeptin, in C and D, most microglia were positive. Unlike the large aggregates attached to cells in serum-free medium, AP in these cells appears to be in cytoplasmic gran- ules. In E and F, AP,_,, was used instead of AP,-,,; little accumulation occurred in spite of the presence of leupeptin. X 300.

198 Ard et al.

Fig. 8. In this electron micrograph of microglia incubated with 5 pg/ml AP with serum and leupeptin, anti-AP immunoposi- tive material is found in membrane-bound intracellular vesi-

cles. The immunogold (15 nm) labeled material is of variable electron density and appears grainy but not obviously fibrillar. X 15,750.

Fig. 9. Microglia cultured with human brain tissue containing Alzheimer’s plaques were only slightly positive for anti-AP immunostaining. A: Phase contrast. Each of the six cultured microglia in the field has been visualized by its content of 3 p m latex beads. The beads were overexposed and therefore are not visible individually. The unfixed cryostat section was incu- bated 5 days in culture so was not well preserved. B: Anti-AP

immunofluorescence. Two microglia (arrows, upper center) in close contact with an immunopositive plaque have taken up little AP and appear mostly dark against the fluorescent plaque. A third microglia (arrow) just to the left of the plaque appears completely dark. Three other microglia (two arrows at left and one at bottom center) are slightly immunopositive; these represent the maximum positive result observed. x 300.

pm latex beads which were applied to the cultures during the final 2 hr preceding fixation. (Beads supplied in 10% suspension were diluted 1: 1,000 in culture medium; in- cubation temperature was 36°C.) Results ranged from completely negative to faintly positive anti-A@ immuno-

staining of microglia (maximum positive immunostain- ing is shown in Fig. 9), regardless of culture conditions. Immunostained plaques were observed in the tissue sec- tions, and microglia were occasionally attached on or near the plaques, but microglia were not concentrated

Scavenging of AD by Microglia 199

Fig. 10. Exogenous AP was accumulated by microglia cul- tured on Alzheimer’s disease brain sections. Six microglia (two together at upper left) are identified by their content of over- exposed 3 pm latex beads in the phase contrast image, A, and by arrows in the immunofluorescent image, B. The four mi- croglia on the right-hand side of the field are immunopositive following incubation with 5 pg/ml exogenous AP for 48 hr and

immunostaining with anti-AP. The two microglia at upper left appear negative, although the tissue around them is immuno- positive. Other deposits of immunofluorescent material, pos- sibly plaques, are present. The bright immunofluorescence of these four positive cells contrasts with the faint positivity ex- hibited by the microglia in Figure 9. X 300.

near the plaques. The same observations apply to blood vessel walls which were occasionally immunostained with anti-Af3 in the sections. From these results it ap- pears that the AP which occurs in plaques in Alzheimer’s brain tissue may be protected from microglial phagocy- tosis in some way compared to purified AP presented in tissue culture medium.

A control experiment demonstrated that microglia plated on cryostat sections retained the ability to accu- mulate AP. Microglia plated on Alzheimer’s brain tissue were incubated in serum-free medium containing 5 pg/ml synthetic AP,-,, for 48 hr. At the end of the incubation, sections were rinsed and latex beads were applied for two hours to label cultured microglia. The cultures were then fixed and immunostained with anti- AP, and intensely immunopositive microglia were found attached to the tissue sections (Fig. lo). This experiment showed that neither endogenous inhibitors nor the vast amount of tissue debris present in the cryostat sections incapacitated microglial accumulation of AP or endocy- tosis of latex beads.

DISCUSSION Microglia consistently appear in the mature (senile

and neuritic) plaques of Alzheimer’s brain (Terry et al., 1964; Itagaki et a]., 1989). It is not known whether they play a role in either depositing or phagocytosing and

degrading amyloid protein in the plaques. This study presents evidence that microglia are capable of endocy- tosis of purified amyloid protein in monomeric or aggre- gate form and that degradation of the protein may occur in some conditions, as evidenced by the effects of lyso- soma1 inhibitors. Surprisingly, amyloid protein in Alz- heimer’s tissue sections was not processed by the cells in the same way as purified AP,-,,.

Microglial Clearance of AP From Serum-Free Medium

Microglial accumulation of AP in short-term cul- tures is readily demonstrated because the undigested ma- terial is present in the cells or on the cell surface in amounts large enough for detection by immunocyto- chemistry or by immunoblotting, even when the amount of AP in the culture medium is low (1-10 p&ml or 0.25-2.5 pM) relative to other tissue culture studies (e.g., Knauer et al., 1992). Immunoblotting shows not only microglial accumulation of AP but also clearance of AP from culture medium.

AP exists in both monomeric (4 kDa) and aggre- gate ( 1 8 kDa and 30 kDa) forms in culture medium in- cubated at 37”C, as shown by immunoblotting. In me- dium incubated with microglia there is a reduction in both monomer and aggregate forms of AP, suggesting that both may be cleared from the medium and taken up by the cells. However, the reduction in monomeric AP

200 Ard et al.

can be explained either by its uptake by the cells, or by its existence in equilibrium with aggregate, so that as aggregate is removed from the medium, monomer is pro- portionally reduced by the continual formation of new aggregate. Of course, aggregates may be formed from monomer by cellular activity, at the cell surface or in- tracellularly. In fact, this is likely since a 30 kDa AP band and a smear of higher molecular weight material are prominent only in the cell layer and not in any other lanes of the immunoblots . Immunostained cultures provide ad- ditional evidence that microglia phagocytose AP aggre- gates, in that immunolabeled material deposited on the culture substratum was usually cleared from the vicinity of individual microglial cells, which were themselves laden with immunopositive inclusions.

Densitometric scanning of immunoblots enables statistical confirmation of microglial clearance of AP us- ing the x2 distribution. The x2 distribution compares the proportion of AP attached to the cell layer and substra- tum to that remaining in the medium in different incu- bation conditions, i.e., with and without microglial cells present. The statistical treatment does not require that the total amount of immunostaining product be equal in the different conditions compared. Absolute amounts of AP cannot be measured in these experiments for several rea- sons: l ) cell lysates were more concentrated than media; 2) immunostaining of the blot was saturated in some areas; 3) antibody may not bind monomer and aggregate with the same affinity. Densitometry is therefore only semi-quantitative. However, this statistic which com- pares the proportionate distribution of immunostaining product is useful given the limitations of the experiment. The significance of microglial removal of AP from cul- ture medium is supported by the x2 test.

Although immunopositive material in microglia is at least partially in higher molecular weight aggregates, it does not appear as fibrils in electron micrographs. Congo red staining of cultured microglia loaded with AP also proved negative. A culture period of 1-2 days, com- bined with the acidity of endosomal compartments in the cells, should be adequate for some AP fibril formation according to in vitro studies (Barrow et al., 1992; Bur- dick et al., 1992; Fraser et al., 1992; Knauer et al., 1992). The absence of fibrillar form in the microglial cultures may reflect interaction of AP with other proteins present in the system.

Microglia cultured in the absence of exogenous AP accumulated little or no immunostainable material intra- cellularly , and immunoblots of such cultures detected no AP either in the cells or in the conditioned medium. It is concluded from this and from the correlation of AP ac- cumulation with concentration of exogenous AP in the medium that the immunostained material accumulated by

than synthesized by the cells. There is an alternative possibility, that exogenous AP stimulates microglia to synthesize P-amyloid precursor protein themselves and then process the precursor to produce AP. However, phagocytosis of sonicated membranes of APP-trans- fected cells or phagocytosis of 3 pm latex beads in cul- ture did not stimulate microglia to accumulate immuno- stainable AP intracellularly (unpublished observations, Ard and Cole), so an hypothesized synthesis of precursor and accumulation of synthesized AP would then be a specific response to exogenous AP, not to phagocytic activity in general. Further evidence on the route of ac- cumulation of immunostainable AP in microglia could be obtained by future studies using iodinated AP (method of Burdick et al., 1992; Knauer et al., 1992), although iodination may alter AP interactions.

In serum-free medium microglia accumulated AP aggregates predominantly at the cell surface, with small amounts of immunostained material found in intracellular vesicles by EM. Apparently the surface-attached material was bound to the membrane, since it was not dislodged by repeated rinsing of cultures during immunostaining. It is not clear why endocytosis of the bound aggregates failed. The cultured microglia were active phagocytes, containing numerous and large vacuoles visible by EM.

Microglial Clearance of A@ in the Presence of Serum

Remarkably, in the presence of fetal bovine serum the disposition of AP-like material in the cultures was entirely different. When AP was added with serum, mi- croglia accumulated little of the protein. With serum and a high concentration of the protease inhibitor leupeptin, cell-associated AP was detectable in intracellular vesi- cles rather than at the cell surface. Because reliable ac- cumulation required protease inhibition, it seems that intracellular proteolysis of AP likely occurred. Alterna- tively, leupeptin could have increased intracellular accu- mulation of AP by increasing uptake of the protein, but such a mechanism of action for leupeptin has not been previously described. It is also possible that leupeptin increased intracellular accumulation of AP by preventing extracellular degradation of it, making more available for uptake. However, the high concentration of leupeptin required argues against an extracellular action in these experiments; it was effective at 100 pg/ml but not at 10 pg/ml. Thus, in some conditions microglia have the ca- pacity to endocytose and degrade A@. In normal brain, despite the production of AP, it does not accumulate in tissue or in CSF, implying the existence of normal clear- ance mechanisms. These may include cell-mediated and extracellular proteases.

A second inhibitor of lysosomal proteolysis, am- the cells was taken up from exogenous sources rather monium chloride, did not significantly enhance AP ac-

Scavenging of AD by Microglia 201

conditions. This tissue, in unfixed cryostat sections, was not denatured by solubilization in SDS and beta-mercap- toethanol, and thereby it differed from the purified plaque cores used in the experiments cited above (Frackowiak et al., 1992; Frautschy et al., 1992).

There are several possible explanations. Possibly AP in tissue sections was complexed with other proteins which either inhibited its uptake by microglia or en- hanced its proteolysis, preventing intracellular accumu- lation of immunostainable AP. If blood-borne macro- phages can phagocytose plaque amyloid in an infarcted area, this demonstrates the capacity of endogenous, non- extracted amyloid to be taken up by phagocytes; how- ever, microglia in Alzheimer’s disease do not appear to be laden with significant plaque amyloid. Another pos- sible explanation for the present results is that microglia on sections may not have access to sufficient AP depos- its. The amount of AP in the sections is difficult to assess since immunostaining is not quantitative, but immunon- egative microglia were found in contact with immunop- ositive plaques. The unfixed tissue certainly presented sectioned surfaces of plaque cores for microglial contact. Finally, it might be questioned whether the endocytotic capacities of microglia were overwhelmed by the amount of tissue debris in the cryostat sections, but in a control experiment microglia on tissue sections did take up AP from culture medium, and latex beads were avidly phagocytosed by microglia in all of these experiments. Thus, of these alternatives, the existence of protein com- plexes which alter AP processing by microglia in tissue seems the most probable.

cumulation by microglia, based on three immunostaining experiments. (A third lysosomal inhibitor, chloroquine, was also tested, but it resulted in early death of the mi- croglia.) In contrast to leupeptin, ammonium chloride does not act directly on proteases, but rather raises the pH of lysosomes; secondarily, it inhibits endosome-ly- sosome fusion (Seglen, 1983). Although these actions inhibit lysosomal proteolysis, they may nevertheless not allow formation of microscopically detectable aggregates of AP within lysosomes. Irnmunostainable accumulation of AP may require fusion of endosomes to increase the concentration of AP within a single vesicle, and it may require the acidic pH of lysosomes to promote aggrega- tion. In this light, it is interesting that AP,p40 was not accumulated from culture medium by microglia, even in the presence of leupeptin. If accumulation depends on formation of protein aggregates, then AP,p,o, which forms aggregates less readily than AP,-,,, would not be expected to accumulate.

The dramatic effect of serum on the disposition of AP may be attributable to the content of potential AB binding proteins in serum. For example, apolipoproteins E and J are known to form complexes with AP, which may alter the pool of free AP in the medium. Lipopro- tein-AP complexes may bind to microglial cell-surface receptors, such as the low density lipoprotein receptor- related protein (LRP), which target ligands for endocy- tosis and lysosomal degradation, whereas free AP aggre- gates may not be endocytosed. Such a difference in processing of free and complexed AP may underlie the reduced endocytosis of AP by microglia cultured in se- rum-free medium.

Microglia Cultured on Tissue Sections The near failure of microglia to accumulate immu-

nostainable AP from human brain tissue containing Alz- heimer’s plaques is surprising in view of the fact that microglia have been shown to phagocytose purified plaque cores both in vitro (Frackowiak et al., 1992) and in vivo when purified amyloid fibrils were injected into rat brain (Frautschy et al., 1992). Their results imply that rat CNS microglia are competent to phagocytose Alzhei- mer’s disease fibrillar amyloid as well as synthetic AP, although in an earlier study it was argued that microglia in Alzheimer’s disease brain do not phagocytose amy- loid fibrils, but that itinerant macrophages associated with infarcts in the Alzheimer’s brain are often laden with phagocytosed amyloid (Wisniewski et al., 199 1). The tissue sections used in the present study contained amy- loid plaques demonstrated both by anti-AP immunostain- ing and by Congo red staining. Because several variations of culture conditions were tested in these experiments, the minimal AP accumulation observed seems to reflect prop- erties of the brain tissue rather than particular culture

SUMMARY The results presented here indicate that cultured

microglia rapidly and effectively remove AP from tissue culture medium and from the surface of the dish and concentrate monomer and aggregates of AP, either in- tracellularly or on the cell surface. Serum proteins radi- cally modify this process. Intriguingly, microglia on Alzheimer’s tissue sections accumulate little AP from the sections, in stark contrast to findings with microglia in rat CNS in vivo or cultured microglia fed synthetic or purified amyloid. The development of an adult CNS mi- croglia culture system to study AP clearance or accumu- lation as aggregates allows us the opportunity to further explore factors, for example in serum or Alzheimer’s disease tissue, which may either promote or inhibit AP removal or deposition and modify the course of P-amy- loidosis in Alzheimer’s disease.

ACKNOWLEDGMENTS The authors wish to thank Dr. John Barker for ad-

vice on statistics, and Brenda Allen, Connie Glasgow,

202 Ard et al.

Glenn Hoskins, Hattie Gatlin, and Ken Mak for technical assistance. This work was supported by an Alzheimer’s Association/F.M. Kirby Foundation Pilot Research Grant (M.D.A.) and the State of California Department of Health Services Alzheimer’s Disease Program (G.M.C.).

REFERENCES Ard MD, Cole GM, Fratkin JD, Wei J (1993): Accumulation of ex-

ogenous Alzheimer’s P amyloid by cultured microglia. SOC Neurosci Abstr 19:185.

Banati RB, Gehrmann J , Czech C , Monning U, Jones LL, Konig G, Beyreuther K, Kreutzberg GW (1993): Early and rapid de novo synthesis of Alzheimer PA4-amyloid precursor protein (APP) in activated microglia. Glia 9: 199-210.

Barrow CJ, Yasuda A, Kenny FTM, Zagorski MG (1992): Solution conformations and aggregational properties of synthetic amyloid P-peptides of Alzheimer’s disease. J Mol Biol225:1075-1093.

Bottenstein JE (1984): Culture methods for growth of neuronal cell lines in defined media. In Barnes DW, Sirbasku DA, Sat0 GH (eds): “Methods for Serum-free Culture of Neuronal and Lym- phoid Cells.” New York: Alan R Liss, pp 3-13.

Burdick D, Soreghan B, Kwon M, Kosrnoski J, Knauer M, Henschen A, Yates J , Cotman C, Glabe C (1992): Assembly and aggre- gation properties of synthetic Alzheimer’s A443 arnyloid pep- tide analogs. J Biol Chem 267546-554.

Busciglio J, Gabuzda DH, Matsudaira P, Yankner BA (1993): Gen- eration of P-amyloid in the secretory pathway in neuronal and nonneuronal cells. Proc Natl Acad Sci IJSA 90:2092-2096.

Frackowiak J, Wisniewski HM, Wegiel J, Merz GS, Iqbal K, Wang KC (1992): Ultrastructure of the microglia that phagocytose amyloid and the microglia that produce P-amyloid fibrils. Acta Neuropathol 84:225-233.

Fraser PE, Nguyen JT, Inouye H, Surewicz WK, Selkoe DJ, Podlisny MB, Kirschner DA (1992): Fibril formation by primate, ro- dent, and Dutch-hemorrhagic analogues of Alzheimer amyloid P-protein. Biochemistry 31: 10716-10723.

Frautschy SA, Cole GM, Baird A (1992): Phagocytosis and deposition of vascular P-amyloid in rat brains injected with Alzheimer P-amyloid. Am J Path01 1401389-1399.

Haass C, Hung AY, Selkoe DJ (1991): Processing of P-amyloid pre- cursor protein in microglia and astrocytes favors an internal localization over constitutive secretion. J Neurosci 11:3783- 3793.

Haass C, Schlossmacher MG, Hung AY, Vigo-Pelfrey C, Mellon A, Ostaszewski BL, Lieberburg I , Koo EH, Schenk D, Teplow DB, Selkoe DJ (1992): Amyloid P-peptide is produced by cul- tured cells during normal metabolism. Nature 359:322-325.

Itagaki S, McGeer PL, Akiyama H, Zhu S, Selkoe D (1989): Rela- tionship of microglia and astrocytes to amyloid deposits of Alzheimer disease. J Neuroimmunol 24: 173-182.

Knauer MF, Soreghan B, Burdick D, Kosmoski J, Glabe CG (1992): Intracellular accumulation and resistance to degradation of the Alzheimer amyloid A443 protein. Proc Natl Acad Sci USA 89:7437-7441,

McCarthy DD, de Vellis J (1980): Preparation of separate astroglial and oligodendroglial cell cultures from rat cerebral tissue. J Cell Biol 85890-902.

McGeer PL, Akiyama H, Kawamata T, Yarnada T, Walker DG, Ishii T (1992): Immunohistochemical localization of beta-amyloid precursor protein sequences in Alzheimer and normal brain tissue by light and electron microscopy. J Neurosci Res 31:

Michler-Stuke A, Wolff JR, Bottenstein JE (1984): Factors influenc- ing astrocyte growth and development in defined media. Int J Dev Neurosci 2575-584.

Schagger H, von Jagow G (1987): Tricine-sodium dodecyl sulfate- polyacrylamide gel electrophoresis for the separation of pro- teins in the range from 1 to 100 kDa. Anal Biochem 166:368- 379.

Scott SA, Johnson SA, Zarow C, Perlmutter LS (1993): Inability to detect P-arnyloid protein precursor mRNA in Alzheimer plaque-associated microglia. Exp Neurol 121: 113-1 18.

Seglen PO (1983): Inhibitors of lysosomal function. Methods Enzy- mol 96:737-764.

Seubert P, Vigo-Pelfrey C, Esch F, Lee M, Dovey H, Davis D, Sinha S , Schlossmacher M, Whaley J, Swindlehurst C, McCormack R, Wolfert R, Selkoe D, Lieberburg I, Schenk D (1992): Iso- lation and quantification of soluble Alzheimer’s 6-peptide from biological fluids. Nature 359:325-327.

Shoji M, Golde TE, Ghiso J, Cheung TT, Estus S, Shaffer LM, Cai X-D, McKay DM, Tinter R, Frangione B, Younkin SG (1992): Production of the Alzheimer amyloid P protein by normal pro- teolytic processing. Science 258: 126-129.

Terry RD, Gonatas NK, Weiss M (1964): Ultrastructural studies in Alzheimer’s presenile dementia. Am J Pathol 44:269-297.

Vaughan DW, Peters A (1981): The structure of neuritic plaques in the cerebral cortex of aged rats. J Neuropathol Exp Neurol40:472- 487.

Wisniewski HM, Wegiel J, Wang KC, Kujawa M, Lach B (1989): Ultrastructural studies of the cells forming amyloid fibers in classical plaques. Can J Neurol Sci 16535-542.

Wisniewski HM, Barcikowska M, Kida E (1991): Phagocytosis of P/A4 amyloid fibrils of the neuritic neocortical plaques. Acta Neuropathol 8 1:588-590.

Wolozin B, Bacic M, Memll MJ, Lesch KP, Chen C, Lebovics RS, Sunderland T ( 1992): Differential expression of carboxyl ter- minal derivatives of amyloid precursor protein among cell lines. J Neurosci Res 33: 163-169.

Wood PM, Bunge RP (1986): Evidence that axons are mitogenic for oligodendrocytes isolated from adult animals. Nature 320:756- 758.

428-442.