Processing of Amyloid Precursor Protein in ... - Wiley Online Library

Journal of Neurochemistry Lippincott—Raven Publishers, Philadelphia © 1997 International Society for Neurochemistry

Processing of Amyloid Precursor Protein in Human Primary Neuron and Astrocyte Cultures *tAndrea C. LeBlanc, *tMaria Papadopoulos, *~c~jj.oljfle Bélair, *tWjlljam Chu, tMilena Crosato, tJaqueline Powell, and ~Cynthia G. Goodyer Departments of *Neurology and Neurosurgery and tPediatrics, McGill University; and tLady Davis Institute for Medical Research, Jewish General Hospital, Montréal, Québec, Canada

Abstract: Increased production of amyloid /3 peptide (A/3) is highly suspected to play a major role in Alzheimer’s disease (AD) pathogenesis. Because A/I deposits in AD senile plaques appear uniquely in the brain and are fairly restricted to humans, we assessed amyloid precursor protein (APP) metabolism in primary cultures of the cell types associated with AD senile plaques: neurons, astrocytes, and microglia. We find that neurons secrete 40% of newly synthesized APP, whereas glia secrete only 10%. Neuronal and astrocytic APP processing generates five C-terminal fragments similar to those observed in human adult brain, of which the most amyloidogenic higher-molecular-weight fragments are more abundant. The level of amyloidogenic 4-kDa A/I exceeds that of nonamyloidogenic 3-kDa A/I in both neurons and astrocytes. In contrast, microglia make more of the smallest C-terminal fragment and no detectable A/I. We conclude that human neurons and astrocytes generate higher levels of amyloidogenic fragments than microglia and favor amyloidogenic processing compared with previously studied culture systems. Therefore, we propose that the higher amyloidogenic processing of APP in neurons and astrocytes, combined with the extended lifespan of individuals, likely promotes AD pathology in aging humans. Key Words: Human primary neuron cultures—Human astrocytes—Human microglia—Amyloid precursor protein—Amyloid /3 peptide—Alzheimer’s disease. J. Neurochem. 68, 1183—1190 (1997).

Amyloid precursor protein (APP) is highly expressed in the brain and undergoes complex metabolism. Intracellular or cell surface APP undergoes asecretase cleavage within the amyloid /3 peptide (A/I) region resulting in the secretion of a 100-kDa nonamyloidogenic N-terminal fragment of APP (sAPP) and a cell-associated 1O-kDa C-terminal fragment. The l0-kDa C-terminal fragment can be further processed into a secreted 3-kDa A/3 fragment (Sambamurti et al., 1992; Haass et a!., 1993, 1995a,b; Kuentzel et a!., 1993; reviewed by LeBlanc, 1994; Thinakaran et a!., 1996). Alternatively, cell surface APP, and possibly

1183

intracellular APP, is processed through the endosomal—lysosomal pathway and generates a series of Cterminal fragments that are degraded in the lysosomes (Golde et a!., 1992; Haass et a!., 1992a; Seubert et a!., 1993; Cheung et a!., 1994). In human brain and neurons, five distinct C-terminal fragments of 8—12 kDa are produced (Estus et al., 1992; LeBlanc, 1995). The 4-kDa A/I, generated in part from endocytosed cell surface APP through proteolytic cleavage of APP by /3- and y-secretases, is normally secreted as a peptide 40 (A/I1-40) or 42/43 (A,81_42) amino acids long. However, intracellular A/I has been detected (Haass et a!., 1992b; Shoji et al., 1992; Busciglio et a!., 1993; Wertkin et a!., 1993; Koo and Squazzo, 1994; LeBlanc and Gambetti, 1994; Martin et a!., 1995). A/I, which is deposited in Alzheimer’ s disease (AD) senile plaques, appears to play a major role in the pathogenesis of AD. It is increased in (a) familial ADassociated mutations of APP or presenilin I (S182) (Citron et a!., 1992, 1994; Cai et a!., 1993; Haass et al., 1994; Nakamura et a!., 1994; Suzuki et a!., 1994; Tamaoka et a!., 1994; Sheuner et al., 1996), (b) transgenic mice overexpressing the APP717 mutation, normal APP, or A/I (Quon et a!., 1991; Games et al., 1995; Higgins et a!., 1995; Hsiao et al., 1995; LaFerla et a!., 1995), and (c) in aging Down’s syndrome individuals who overexpress APP and A/I possibly as a result of increased gene dosage (Wisniewski, et a!., 1985; Tanzi et al., 1987; Teller et a!., 1996). The pathological manifestations of AD are fairly restricted to primates and are confined to the brain. Yet, investigations on APP processing are limited in neuronal cells of human origin. Select studies show Received September 4, 1996; revised manuscript received October 29, 1996; accepted November 5, 1996. Address correspondence and reprint requests to Dr. A. C. LeBlanc at Lady Davis Institute, 3755 Ch. Côte Ste-Catherine, Montréal, Québec, Canada H3T 1E2. Abbreviations used: A/3, amyloid /3 peptide; AD, Alzheimer’s disease; APP, amyloid precursor protein; KPI, Kunitz protease inhibitor; sAPP, a-secretase-cleaved secreted amyloid precursor protein.

A. C. LEBLANC ET AL.

1184

higher !eve!s of 4-kDa than 3-kDa A/I in human fetal mixed brain cultures, iso!ated human astrocytes, neurona!!y differentiated human teratocarcinoma NT2 ce!!s (NT2N), and human APP-infected rat primary cultures in contrast to higher levels of 3-kDa A/I in most other ce!! !ines (Buscig!io et a!., 1993; Wertkin et a!., 1993; Simons et a!., 1996). Rodent primary neuron and astrocyte cultures express high levels of APP in a ce!l-specific manner (LeB!anc et a!., 1991). These neurons express mainly APP695, whereas g!ia! ce!!s generate mostly Kunitz protease inhibitor (KPI) containing APPs: APP751 and APP770. Although expressed at high leve!s, litt!e APP is processed through the amyloidogenic pathways in the rodent neurons or astrocytes. Higher levels of 3-kDa than 4-kDa A/I are produced, and C-termina! fragments are difficult to detect (LeBlanc et al., 1996). Because rodents fail to develop AD patho!ogy, we undertook a study of APP expression and metabolism in primary human fetal neuron, astrocyte, and microg!ia! cultures. The results of these experiments show a high degree of amyloidogenic processing in both human neurons and astrocytes but not microglia cu!tures. In addition, the human neurons and astrocytes produce more amyloidogenic APP fragments than rodent primary brain cultures, nonneuronal primary cultures, and nonneurona! and most neurona! continuous cultures. MATERIALS AND METHODS Cultures The brains were collected from human fetuses at the time of the therapeutic abortion in accordance to the Quebec Health Code and under IRB approval. Human primary neurons were prepared as previously described (LeB!anc, 1995) from fetal brains of 12—17 weeks (Munsick, 1984). In brief, the cells were dissociated with trypsin and deoxyribonuclease I and plated in 5% decomplemented serum in highglucose-containing minima! essential medium with Earle’s salts, 1 mM sodium pyruvate, and 2 mM glutamine. Proliferating ce!ls were inhibited with 1 mM fluorodeoxyuridine. Astrocytes were purified by vigorous shaking of the flasks 3 days after plating dissociated brain cells (LeBlanc et al., 1991). Mic oglia were detached by gentle shaking of mixed cultures grown for 3 weeks without changing the media. The analysis was carried out on neurons at 10 days in culture, whereas astrocytes and microglia required 3 weeks before enough cells were obtained for the experiments. In each experiment, the cultures of neurons, astrocytes, and microglia were monitored for percent purity by immunocytochemistry as described previously (LeBlanc, 1995). Neuron, astrocyte, and microglia cultures were obtained at ~ 99, and 100% purity, respectively.

Characterization of APP mRNA by reverse transcription-PCR RNA was obtained by guanidinium-phenol extraction (Chomczynski and Sacchi, 1987). Single-strand cDNA was made using random primers as described (LeBlanc et al., 1989). Forward primer 5’ AAG AGG TGG TTC GAG 3’ (nucleotides 851—865 of APP751 starting at the ATG codon)

J. Neurochem., Vol. 68, No. 3, 1997

and reverse primer 5’ CAT GTC GGA ATT CTG C 3’ (nucleotides 1,975—1,960 of APP751) were designed to amplify A/I-containing cDNAs and to identify the three major differentially spliced APP mRNA species: APP695, APP751, and APP770. The APP eDNA was amplified for 30 cycles as follows: 1 mm of 94°Cdenaturation, 1 mm of 60°Cannealing, and 1 mm of 72°Celongation. PCR products were separated on a polyacrylamide gel using the GeneAmp Detection Gel system (Perkin Elmer, Montreal, Quebec, Canada).

Metabolic labeling of cells and immunoprecipitations for Cells were rinsed and incubated for 135S]methionine h in methionineadding 250 jtCi/ml [ After labeling, the 5free h ormedium for I h before in pulse—chase experiments. cells were lysed in Nonidet P-40 lysis buffer (1% Nonidet P40, 5 mM EDTA, 0.05% phenylmethylsulfonyl fluoride, 0.1 ~.tg/ml pepstatin A, I ~tg/ml N-a-p-tosyl-L-lysine chloromethyl ketone, and 0.5 ~ig/ml leupeptin), and the detergent-insoluble fraction was separated by centrifugation. The detergent-insoluble pellet was resolubilized in 70% formic acid or l,l,1,3,3,3-hexafluoro-2-propanol (Burdick et al., 1992) and tested for the presence of APP metabolicproducts. Because none was found, the pellet was discarded in subsequent experiments. Immunoprecipitations of cellular APP and sAPP were carried out in IX RIPA (50 mM Tris, pH 8.0, 150 mM NaC!, 1% Nonidet P-40, and 5 mM EDTA) (Lane, 1989). Full-length APP was immunoprecipitated withanti-C 21 (LeBlanc and Gambetti, 1994; LeBlanc, 1995), anti-N (a kind gift from B. Greenberg, Cephalon), 4G8 (against A/I17_24), and 6E10 (against A/I116) (Kim et al., 1 990a, b), whereas sAPP was immunoprecipitated with antiN and 6E10. Immunoprecipitated APP and APP metabolic products were quantified by phosphorimaging (Molecular Dynamics). We corrected the quantitative data for the amount of methionine present in the APP or APP fragments as deduced from the APP sequence or previously published sequences of the fragments (Cheung et al., 1994; Simons et al., 1996). Neuronal cellular APP695 contains 21 methionines, whereas astrocytic KPI-APPs contain 23. sAPP lacks four methionines compared with cellular APP. C-terminal fragments CO and Cl, C2—C4, and CS contain five, four, and three methionines, respectively. A/I contains one methionine. The percent sAPP was calculated as (no. of pixels of sAPP at the time cellular APP disappeared in a pulse—chase! no. of pixels of cellular APP holoprotein immunoprecipitated with anti-N at time 0 of chase) X 100. Both sAPP bands were counted for radioactivity in neurons and microglia. The percent 4-kDa A/I per total amyloid peptide was calculated as (no. of pixels of 4-kDa A/I/no, of pixels of 4-kDa + 3kDa A/I) X 100. The percent 4-kDa A/I/APP holoprotein was calculated as (no. of pixels of 4-kDa A/I/no, of pixels of cellular APP holoproteins immunoprecipitated with antiN or anti-C21) X 100. Each experiment was conducted on at least three independent cultures. Statistical differences were determined by two-tailed unpaired t tests.

RESULTS APP mRNA is differentially spliced into three major A/I-containing mRNAs: APP695, APP751, and APP770. We developed a reverse transcription-PCR system to measure A/I-containing APP mRNA levels and oh-

APP PROCESSING IN HUMAN PRIMARY NEURONS

1185

served that APP

695 is the most highly expressed APP mRNA in human neurons, whereas APP751 is most abundant in astrocytes (data not shown). APP770 is absent in neurons and expressed at low levels in astrocytes. The APP mRNA was not determined in microg-

ha because of limited availability of cells, but APP production was determined by metabolic labeling as shown below. Cellular APP holoprotein production in primary neuron, astrocyte, and microglia cultures Cellular APP synthesis was monitored by metabolic labeling and followed by immunoprecipitations with various antibodies directed against APP (Fig. 1A). Immunoprecipitation of full-length cellular APP (or APP holoprotein) with the APP C-terminal and Nterminal antisera, anti-C21 and anti-N, respectively, showed that the APP isoforms varied considerably in each cell type. In neurons, antibodies to the N-terminal region of APP immunoprecipitated four bands in the expected size range of 95—110 kDa for APP695. Antisera to the C-terminal region of APP (C21) immunoprecipitated the three upper bands but not the lower one, indicating that the lower band has a truncated C-terminus similar to that reported in rat neurons and identified as APPL~C(LeBlanc et a!., 1996). The C-terminal immunoprecipitations were comp!ete!y competed out with synthetic C21 peptide (C21 +). Immunoprecipitation of the three upper bands with 4G8, which recognizes amino acids 17—24 of A/I, confirmed that these are APPs and not APP-like proteins (Wasco et al., 1993). Compared with neurons, astrocytes synthesize APPs of higher molecular weight, as expected for the KPIcontaining isoforms, APP751 and APP770. Four proteins are immunoprecipitated with the anti-C21 antiserum and completely competed out with synthetic peptide, but only the three highest-molecular-weight species are also immunoprecipitated with anti-N, 4G8, and 6E10 (epitopes detected by these antibodies are depicted in the schematic diagam of Fig. 1A). The low levels of APP immunoprecipitated with 4G8 and 6E10 are due to the lower avidity of these antibodies in the immunoprecipitation of APP rather than demonstrate a higher amount of APP-like proteins. On immunoblots, 6E1 0 detects APP in similar amounts as anti-N (data not shown). In microghia, as in astrocytes, three APPs are immunoprecipitated with anti-C21, anti-N, and 4G8. Also, a fourth protein migrating under these three is additionally immunoprecipitated with anti-C21 and could represent an APP-like protein. Secretion of APP by primary neuron, astrocyte, and microglia cultures sAPP, resulting from the secretory APP metabolic pathway, is easily detected in the medium from neuron, astrocyte, and microglia cultures (Fig. lA). Unexpectedly, two distinct APPs are secreted by neurons and

FIG. 1. APP expression in human brain, neurons, astrocytes, and microglia. A: Epitopes detected by various antibodies to APP and APP production in primary neuron, astrocyte, and microglial cultures. Schematic diagram shows the epitopes detected by various APP antibodies, the sites of a-, /3-, and ymethionine-labeled APP and 35S] secretase cleavage within the A/3 region, and cellular autoradiograms sAPP from neuron, astrocyte, and microglia cultures. C21 + repofimmunoprecipitated [ resents a competition of the anti-C 21 immunoprecipitation by addition of C21 peptide. B: Production of sAPP after APP metabolism in neurons, astrocytes, and microglia. The cells were radiolabeled for 60 mm and chased in complete unlabeled media. sAPP and cellular APP were immunoprecipitated with anti-N from the media and cell lysate, respectively. The percent sAPP represents the amount of sAPP at the time radiolabeled cellular APP has been chased out (480 mm) divided by the amount of cellular APP at 0 mm of chase. All three cell types assayed were cultured from the same brain, and data are mean ± SD (bars) values of three cultures from different brain samples. Neurons versus astrocytes, p = 0.03; astrocytes versus microglia, p = 0.4.

microghia, whereas only one is detected from astrocytes. In neurons, the two sAPPs are recognized by

6ElO but not by 4G8, indicating that they result from the normal a-secretase cleavage and are likely the result of variable posttranslational modification as re-

ported in rodent neurons (Simons et a!., 1996). sAPP was quantified when all cellular APP had been chased out (usually 480 mm), and its level was expressed as the percentage of APP secreted from the

initially labeled cellular APP (time 0 mm of chase) (Fig. lB). The amount of newly synthesized APP secreted is fourfold higher in neurons than in astrocytes

or microghia (p

=

0.03).


1186


cell types. Previous characterization of the neuronal

C-terminal fragments showed that Cl —C4 are also immunoprecipitated with 4G8, indicating the presence of amino acids 17—24 of A/I in these fragments (LeBlanc, 1995). Based on the immunoprecipitation results and size, it is likely that CO and Cl contain the entire A/I sequence. The amount of C-terminal fragments differs considerably in each cell type. In neurons, C2 is the predominant C-terminal fragment followed closely by C4 and C5 (Fig. 2B); Cl and C3 are the least abundant. In astrocytes, CO is the most abundant by far and is fol-

lowed by C4, with lesser amounts of C3, Cl, and C2 and barely detectable CS. In contrast, C4 or C5 is the predominant band observed in microghia (Fig. 2A). As the microghial C-terminal fragments were minimal,

we did not quantify the amount as was done in astrocytes and neurons. To determine if neurons metabolized more of their APP through the endosomal—lysosomal pathway, we expressed the total amount of C-terminal fragments as a percentage of the amount of APP holoprotein immunoprecipitated with anti-C21. We find that neu-

rons generate more C-terminal fragments per APP holoprotemn than astrocytes and microglia (Fig. 2C). However, the amount of C-terminal fragments produced varies considerably in different neuronal preparations as indicated by the wide SD (n = 4). FIG. 2. Production of C-terminal fragments in neurons, astrocytes,35S]methionmne, and microglia. Each culture fragments was metabolically labeled and cell C-terminal were immunowith [ precipitated with anti-C 21. A: Autoradiogram shows immunoprecipitated C-terminal fragments from neurons, astrocytes, and microglia. Astrocytes (+) represents competition of the immunoprecipitation of C-terminal fragments in astrocytes by synthetic C21 peptide. B: Percentage of each C-terminal fragment relative to total amount of C-terminal fragments immunoprecipitated with anti-C21 in neurons and astrocytes. Data are mean ±SD(bars) values of three independent experiments. C: Proportion of the total C-terminal fragments relative to full-length APP immunoprecipitated with anti-C21 in neurons, astrocytes, and microglia. Data are mean ±SD (bars) values ofthree independent cultures. No statistical difference is obtained among the three groups.

Endosomal—lysosomal processing of APP in neurons, astrocytes, and microglia Five C-terminal fragments (C1—C5, from largest to smallest), simihar to those observed in human brain, are immunoprecipitated with anti-C21 in neuron, astrocyte, and microghia primary cultures (Fig. 2A) (Estus eta!., 1992; LeBlanc, 1995). A sixth C-terminal fragment (CO) migrating above Cl and competed with C21 synthetic peptide is observed in astrocytes and microglia. The top band present in all lanes of Fig. 2A is a nonspecific protein because it cannot be competed out with C21 synthetic peptide but serves as an idea! marker to align C-terminal fragments from different


A~expression in neurons and astrocytes Two peptides of 3 kDa (sometimes as a doublet) and 4 kDa are immunoprecipitated from neuron and astrocyte culture media with the 4G8 antibody (Fig. 3A), whereas no detectable A/I was ever found in microglial cultures (data not shown). The 4-kDa A/I represents either A/I140 or A/I142143. This A/I is also immunoprecipitated with 6E1O and competed out with synthetic A/I140 peptide, as shown previously (LeB!anc, 1995). The 3-kDa band likely represents the nonamyhoidogenic 3-kDa A/I or p3 (Haass et a!.,

1993). To determine the level of amyloidogenic A/I compared with nonamyloidogenic A/I, the amount of 4-kDa A/I was expressed relative to the total amount of A/Is secreted (3 + 4 kDa). The 4-kDa A/I represents almost 50% of the total A/Is secreted from the neuron cultures, which is slightly, but not significantly, higher than that generated by astrocytes (p = 0.6) (Fig. 3B). To determine if the amount of 4-kDa A/I

produced from the APP is higher in neurons than in astrocytes, the amount of 4-kDa A/I was expressed

relative to the amount of APP holoprotein immunoprecipitated with the anti-C21 antisera. The results show that the production of 4-kDa A/I is higher in astrocytes

than in neurons (p

=

0.06) (Fig. 2C).

DISCUSSION It is believed that overexpression or abnormal processing of APP leads to an overproduction of the amy-

APP PROCESSING IN HUMAN PRIMARY NEURONS

FIG. 3. Production ofA/I in neurons and astrocytes. A: Autoradiogram shows the immunoprecipitation of A/I from neurons and astrocytes with 4G8. B: Percent A/I secreted as a 4-kDa peptide in neurons and astrocytes compared with total A/I. Data are mean ± SD (bars) values of three experiments. p < 0.6. C: Percent 4-kDa A/I secreted relative to APP holoprotein in neurons and astrocytes. Data are mean ±SD (bars) values of three experiments. p < 0.06.

loidogenic 4-kDa A/I in AD brains. Studies of APP metabolism indicate that amyloidogenic processing through the endosoma!—!ysosomal pathway, which generates either the C-terminal fragments or 4-kDa A/I, is minimal in human nonneurona! primary cultures

and in nonneurona! and most neuronal human continuous cell lines even when transfected to overexpress the human APP (Palmert et a!., 1989; Go!de et a!., 1992; Haass et a!., 1992a,b; Shoji et a!., 1992). Similarly, neonatal rat primary neuron and astrocyte cu!tures, which naturally express high !eve!s of APP695 or KPIcontaining APPs, respectively, do not generate significant levels of C-terminal fragments and 4-kDa A/I (LeBlanc et a!., 1996). Because AD pathology is fairly

restricted to the brain of primates, we hypothesized that possibly APP metabohism in nonprimate neural and peripheral cells or primate peripheral cells is less amyloidogenic compared with APP processing in the human brain cell types. Therefore, in the present study, we assessed APP metabolism in the major human cell types associated with the senile plaques: neurons, astrocytes, and microghia. We find that both human neurons and astrocytes generate high levels of amyloidogenie fragments compared with human microglia. The amyloidogenic fragments are also more abundant in

1187

human neurons and astrocytes than in rat primary neuron and astrocyte cultures or transfected nonneuronal and most neurona! ce!l hines. Both human primary neurons and astrocytes, but not microghia, show higher levels of 4-kDa than 3-kDa A/I as previously shown in mixed human fetal brain and astrocyte cultures and in the NT2N neuronal cell hine (Buscighio et a!., 1993; Wertkin eta!., 1993). Compared with rat primary cultures, human neurons and astrocytes secrete 22 and 100 times more 4-kDa A/I, respectively (LeBlanc et a!., 1996). We sustain that these human neuron, astrocyte, and microghial primary cultures are the most feasib!e ce!!ular models to understand APP metabolism in human brain. The use of fetal rather than adult eel! types is of course not ideal because metabolism may change during aging. However, the identica! pattern of C-terminal fragments between human primary neuron cu!tures and adult human brain (Estus et a!., 1992) indicates that at least the C-terminal fragment production is quantitatively identical in fetal and adult tissue. To our knowledge, these five C-terminal fragments have not been identified in other cellular models of human APP metabolism. Rat hippocampa! neurons infected with human APP yield only four of these C-terminal fragments, similar to those fragments identified in the transfected human Ml7 neuroblastoma cell line (Cheung et al., 1994; Simons et al., 1996). Our results also show that not only the type of C-terminal fragments but also the proportion of each is similar to that of human adult brain. The higher-mo!ecu!ar-weight Cterminal fragments are rarely as abundant as seen in these human cultures and brain tissue. The fact that these higher-molecular-weight fragments are also more abundant in the human APP-infected rodent primary neurona! cultures indicates that processing is dependent on either the human sequence or increased expression of APP (Simons et a!., 1996). It has been proposed that the po!arity of neurons may dictate the metabolism of APP, thereby resu!ting in increased amyloidogenic processing. An analogy can be made between the baso!atera! and apical compartment of the polarized epithehial Madin-Darby canine kidney cel!s and the dendritic and axonal compartments of neurons, respectively (Huber et a!., 1993). Approximately 90% of APP is present on the eel! surface of the baso!atera! compartment in Madin-Darby canine kidney cells, and sAPP, 4-kDa A/I, and 3-kDa A/I are secreted into the basolatera! compartment (Haass et a!., l995a; Yamazaki a!., missorted 1995). The 1e mutant APP is et partly to “Swedish” dout. the apical compartment of Madin-Darby canine kidney cel!s (De Strooper et a!., 1995). Because the “Swedish” APP generates an eight- to 15-fo!d increase in 4kDa A/I (Citron et al., 1992; Cai et a!., 1993), it is possible that polarized sorting affects the level of A/I produced in Madin-Darby canine kidney cells. Studying APP processing in neurons and astrocytes allows


1188


a comparison of APP metabolism in polar and nonpolar brain cell types expressing high levels of endogenous APP. Cultured astrocytes are considered reactive (Ma!hotra et a!., 1990) but in our cultures maintain the “fibrob!ast-!ike” flat shape of a nonpolar cell. We find that the amount of sAPP secreted is fourfold higher in neurons than in astrocytes and microghia. It is unlikely that secretion is mediated through polarized sorting because rat astrocytes secrete more sAPP than rat neurons (LeBlanc et a!., 1996). Astrocytes and neurons

both make high levels of C-terminal fragments and 4kDa A/I, indicating that metabolism of APP through the amyloidogenic pathways is not necessarily different in polar and nonpolar cells. There are many other possible factors that could regulate APP metabolism

in human neurons and astrocytes compared with other cell types, such as (a) elevated levels or activities of /3- and y-secretases, (b) optimized trafficking of APP through amyloidogenic pathways, or (c) intrinsic properties of the human APP sequence. We have been concerned with the possible variabi!ity in the data because of the inevitable diverse genetic

background of the brain tissue. However, we find that the results are most consistent in independent cell cu!tures. We do not observe any difference in the APP metabolism profile of cultures obtained from brain tissues between 12 to 17 weeks of development, indicating that the age of the tissue at least within this time frame does not directly affect the metabolism of APP. We have now repeated these experiments in at least 10 more neuronal preparations as controls in studies of the regulation of APP processing, and the results are consistently the same as shown in this study. In conclusion, we believe that at least part of the reason that senile plaques are mostly restricted to the primate brain could lie within the cell-specific APP

metabolism observed in human neurons and astrocytes, two cell types physically associated with senile plaques. Normal brains contain mostly resting astrocytes, which normally express low levels of APP. Therefore, resting astrocytes are unlikely to contribute significantly to the pool of amyloidogenic peptides in

the normal brain. However, APP expression is strongly increased in injury models where extensive gliosis occurs (Siman et al., 1989). Gliosis is often a predomi-

nant feature of AD pathology, and reactive astrocytes are usually associated with senile plaques. Our results indicate that, in their reactive state, astrocytes could make a significant contribution to the soluble poo1 of A/I in human AD brain. Microghia, which are also intimately linked to the plaque, are highly unlikely to contribute to the production of amyloidogenic fragments. In view of the more amyloidogenic processing in the human neurons and astrocytes, it is reasonable to speculate that the familial AD APP mutations, which have been shown to increase the levels of 4-kDa A/I in transfected nonneuronal cultures, exacerbate the already high production of 4-kDa A/I and C-terminal


fragments in these patients, thereby resulting in amyloidogenicity. Acknowledgment: The authors express sincere thanks to Barry Greenberg

Hemant Paudel (McGill University) and

(Cephalon) for the kind gift of T and anti-N APP antisera, respectively. This research was supported by grant ROl

NS3 1700 from the National Institutes of Health, the Alzheimer Society of Canada, and Fonds de Recherche en Sante du

Québec to A.C.L. and by funds from the Medical Research Council to C.G.G. REFERENCES Burdick D., Soreghan B., Kwon M., Kosmoski J., Knauer M., Henschen A., Yates J., Cotman C., and Glabe C. (1992) Assembly and aggregation properties of synthetic Alzheimer’s A4//I amyloid peptide analogs. J. Biol. Chem. 267, 546—554. Busciglio J., Gabuzda D., Matsudaira P., and Yankner B. (1993) Generation of /3-amyloid in the secretory pathway in neuronal and nonneuronal cells. Proc. Nati. Acad. Sci. USA 90, 2092— 2096. Cai X., Golde T., and Younkin S. (1993) Release of excess amyloid /3 protein from a mutant amyloid /3 protein precursor. Science 259, 514—516. Cheung T., Ghiso J., Shoji M., Cai X., Golde T., Gandy S., Frangione B., and Younkin S. (1994) Characterization by radiosequencing of the carboxyl-terminal derivatives produced from normal and mutant amyloid /3 protein precursors. Amyloid mt. J. Exp. Gun. Invest. 1, 30—38. Chomczynski P. and Sacchi N. (1987) Single-step method of RNA isolation by acid guanidinium thiocyanate-phenol-chloroform extraction. Anal. Biochem. 162, 156—159. Citron M., Oltersdorf T., Haass C., McConlogue L., Hung A., Seubert P., Vigo-Pelfrey C., Lieberburg 1., and Selkoe D. (1992) Mutation of the /3-amyloid precursor protein in familial Alzheimer’s disease increases /3-amyloid protein production. Nature 360, 672—679. Citron M., Vigo-Pelfrey C., Teplow D., Miller C., Schenk D., Johnston J., Winblad B., Venizelos N., Lannfelt L., and Selkoe D. (1994) Excessive production of amyloid /3-protein by peripheral cells of symptomatic and presymptomatic patients carrying the Swedish familial Alzheimer disease mutation. Proc. Nail. Acad. Sci. USA 91, 11993—11997. De Strooper B., Craessaerts K., Dewachter I., Moechars D. G. B., Van Leuvan F., and Van Den Berghe H. (1995) Basolateral secretion of amyloid precursor protein in Mardin—Darby canine kidney cells is disturbed by alterations of intracellular pH and by introducing a mutation associated with familial Alzheimer’s disease. J. Biol. Chem. 270, 4058—4065. Estus S., Golde T., Kunishita T., Blades D., Lowery D., Eisen M., Usiak M., Qu X., Tabira T., Greenberg B., and Younkin S. (1992) Potentially amyloidogenic, carboxy-terminal derivatives of the amyloid protein precursor. Science 255, 726—728. Games D., Adams D., Alessandrini R., Barbour R., Berthelette P., Blackwell C., Can T., Clemens J., Donaldson T., Gillespie F., Guido T., Hagopian S., Johnson-Wood K., Khan K., Lee M., Leibowitz P., Lieberburg I., Little S., Masliah E., McConlogue L., Montoya-Zavala M., Mucke L., Paganini L., Penniman E., Power M., Schenk D., Seubert P., Snyder B., Soriano F., Tan H., Vitale J., Wadsworth S., Wolozin B., and Zhao J. (1995) Alzheimer-type neuropathology in transgenic mice overexpressing V717F /3-amyloid protein. Nature 373, 523—527. Golde T., Estus S., Younkin L., Selkoe D., and Younkin S. (1992) Processing of the amyloid protein precursor to potentially amyloidogenic derivatives. Science 255, 728—730. Haass C., Koo E., Mellon A., Hung A., and Selkoe D. (l992a) Targeting of cell-surface /3-amyloid precursor protein to lyso-

APP PROCESSING IN HUMAN PRIMARY NEURONS somes: alternative processing into amyloid-bearing fragments. Nature 357, 500—503. Haass C., Schlossmacher M., Hung A., Vigo-Pelfrey C., Mellon A., Ostaszewski B., Lieberburg I., Koo E., Schenk D., Teplow D., and Selkoe D. (1992b) Amyloid /I-peptide is produced by cultured cells during normal metabolism. Nature 359, 322—325. Haass C., Hung A., Schlossmacher M., Teplow D., and Selkoe D. (1993) /3-Amyloid peptide and a 3-kDa fragment are derived by distinct cellular mechanisms. J. Biol. Chem. 268, 3021— 3024. Haass C., Hung A., Selkoe D., and Teplow D. (1994) Mutations associated with a locus for familial Alzheimer’ s disease result in alternative processing of amyloid /3-protein precursor. J. Biol. Chem. 269, 17741—17748. Haass C., Koo E. H., Capell A., Teplow D., and Selkoe D. (1995a) Polarized sorting of /I-amyloidprecursor protein and its proteolytic products in MDCK cells is regulated by two independent signals. J. Cell Biol. 128, 537—547. Haass C., Lemere C., Capell A., Citron M., Seubert P., Schenk D., Lannfelt D., and Selkoe D. (1995b) The Swedish mutation causes early-onset Alzheimer’s disease by /I-secretase cleavage within the secretory pathway. Nature Med. 1, 1291—1296. Higgins L. S., Rodems J. M., Catalano R., and Cordell B. (1995) Early Alzheimer disease-like histopathology increases in frequency with age in mice transgenic for /I-APP75 1. Proc. NatI. Acad. Sci. USA 92, 4402—4406. Hsiao K., Borcheldt D., Olson K., Johannsdottir R., Kitt C., Yunis W., Xu S., Eckman C., Younkin S., Price D., ladecola C., Clark

B., and Carlson G. (1995) Age-related CNS disorder and early death in transgenic FVB/N mice overexpressing Alzheimer amyloid precursor protein. Neuron 15, 1203—1208. Huber L. A., de Hoop M. J., Dupree P., Zerial M., Simons K., and Dotti C. (1993) Protein transport to the dendritic plasma membrane of cultured neurons is regulated by rab8p. J. Cell Biol. 123, 47—55. Kim K., Miller D., Sapienza V., Chen C., Bai C., Grundke-Iqbal I., Currie J., and Wisniewski H. (1990a) Production and characterization of monoclonal antibodies reactive to synthetic cerebrovascular amyloid peptide. Neurosci. Res. Common. 2, 121—130. Kim K. S., Wen G. Y., Bancher C., Chen C. M. J., Sapienza V. J., Hong H., and Wisniewski H. M. (1990b) Detection and quantitation of amyloid /I-peptide with 2 monoclonal antibodies. Neurosci. Res. Commun. 7, 113—122. Koo E. H. and Squazzo S. L. (1994) Evidence that production of amyloid /3-protein involves the endocytotic pathway. J. Biol. Chem. 269, 17386—17389. Kuentzel C., Caporaso G., Thyberg J., Gandy S., and Greengard P. (1993) The Alzheimer /3-amyloid protein precursor/protease nexin-Il cleaved by secretase in a trans-Golgi secretory compartment in human neuroglioma cells. Biochem. J. 295, 367— 378. LaFerla F. M., Tinkle B. T., Bieberich C. J., Haudenschild C. C., and Jay G. (1995) The Alzheimer’s A/I peptide induces neurodegeneration and apoptotic cell death in transgenic mice. Nature Genet. 9, 21—30. Lane D. (1989) Immunoprecipitation, in Antibodies (Harlow L., ed), pp. 421—470. Cold Spring Harbor Laboratory, Cold Spring Harbor, New York. LeBlanc A. C. (1994) The role of /3-amyloid peptide in Alzheimer’s disease. Metab. Brain Dis. 9, 3—31. LeBlanc A. C. (1995) Increased production of 4 kDa amyloid /3 peptide in serum deprived human primary neuron cultures: possible involvement of apoptosis. J. Neurosci. 15, 7837—7846. LeBlanc A. and Gambetti P. (1994) Production of Alzheimer 4kDa /I-amyloid peptide requires the C-terminal cytosolic domain of the amyloid precursor protein. Biochem. Biophys. Res. Common. 204, 1371—1380. LeBlanc A., Foldvari M., Spencer D., Breckenridge W., Fenwick R., Williams D., and Mezei C. (1989) The apolipoprotein A-

1189

1 gene is actively expressed in the rapidly myelinating avian peripheral nerve. J. Cell Biol. 109, 1245—1256. LeB lane A., Chen H., Autilio-Gambetti L., and Gambetti P. (1991) Differential APP gene expression in neonatal rat cerebral cortex, meninges and primary astroglial, neuronal and microglial cells. FEBSLett. 292, 171—178. LeBlanc A. C., Xue R., and Gambetti P. (1996) Amyloid precursor protein metabolism in primary cell cultures of neurons, astrocytes, and microglia. J. Neurochem. 66, 2300—23 10. Malhotra S. K., Shnita T. K., and Elbrink J. (1990) Reactive astrocytes—a review. Cytobios 61, 133—160. Martin B. L., Schrader-Fisher G., Busciglio J., Duke M., Paganetti P., and Yankner B. A. (1995) Intracellular accumulation of/Iamyloid in cells expressing the Swedish mutant amyloid precursor protein. J. Biol. Chem. 270, 26727—26730. Munsick R. A. (1984) Human foetal extremity lengths in the interval from 9 to 21 menstrual weeks of pregnancy. Am. J. Obstet. Gynecol. 149, 883—887. Nakamura T., Shoji M., Harigaya Y., Watanabe M., Hosoda K., Cheung T., Shaffer L., Golde T., Younkin L., Younkin S., and Hirai S. (1994) Amyloid /3 protein levels in cerebrovascular fluid are elevated in early-onset Alzheimer’s disease. Ann. Neurol. 36, 903—911. Palmert M., Podlinsny M., Witker D., Oltersdorf T., Younkin L., Selkoe D., and Younkin S. (1989) The /3-amyloid protein precursor of Alzheimer disease has soluble derivatives found in human brain and cerebrospinal fluid. Proc. Natl. Acad. Sci. USA 86, 6338—6342. Quon D., Wang Y., Catalano R., Scardina J., Murakami K., and Cordell B. (1991) Formation of /3-amyloid protein deposits in brains of transgenic mice. Nature 352, 239—241. Sambamurti K., Shioi J., Anderson M. A., Pappolla M. A., and Robakis N. K. (1992) Evidence for intracellular cleavage of the Alzheimer’s amyloid precursor in PCI 2 cells. J. Neurosci. Res.

33, 319—329. Seubert P., Oltersdorf T., Lee M., Barbour R., Blomquist C., Davis D., Bryant K., Fritz L. C., Galasko D., Thal L. J., Lieberburg I., and Schenk D. (1993) Secretion of /3-amyloid precursor protein cleaved at the amino terminus of the /I-amyloid peptide. Nature 361, 260—263. Sheuner D., Eckman C., Jensen M., Song X., Citron M., Suzuki N., Bird T., Hardy J., Hutton M., Kukull W., Larson E., LevyLahad E., Viitanen M., Peskind E., Poorkaj P., Schellenberg G., Tanzi R., Wasco W., Lannfelt L., Selkoe D., and Younkin S. (1996) Secreted amyloid /3-protein similar to that in the senile plaques of Alzheimer’s disease is increased in vivo by the presenilin 1 and 2 and APP mutations linked to familial Alzheimer’s disease. Nature Med. 2, 864—870. Shoji M., Golde T., Ghiso J., Cheung T., Estus S., Shaffer L., Cai X., McKay D., Tintner R., Frangione B., and Younkin S. (1992) Production of the Alzheimer amyloid /3 protein by normal proteolytic processing. Science 258, 126—129. Siman R., Card P., Nelson R., and Davis L. (1989) Expression of /I-amyloid precursor protein in reactive astrocytes following neuronal damage. Neuron 3, 275—285. Simons M., DeStrooper B., Multhaup G., Tienari P. J., Dotti C. G., and Beyreuther K. (1996) Amyloidogenic processing of the human amyloid precursor protein in primary cultures of rat hippocampal neurons. .1. Neurosci. 16, 899—908. Suzuki N., Cheung T., Cai D., Odaka A., Laslo 0., Eckman C., Golde T., and Younkin S. (1994) An increased percentage of long amyloid /3 protein secreted by familial amyloid /3 protein precursor (/IAPP1 17) mutants. Science 264, 1336—1340. Tamaoka A., Okada A., Ishibashi Y., Usami M., Sahara N., Suzuki N., Nukina N., Mizusawa H., Shoji S., Kanazawa I., and Mori H. (1994) APP717 missense mutation affects the ratio of amy-

bid /3 protein species (1—42/43 and 1—40) in familial Alzheimer’s disease brain. J. Biol. Chem. 269, 32721—32724. Tanzi R., Gusella J., Watkins P., Bruns G., St. George-Hyslop P., Van Keuren M., Patterson D., Pagan S., Kumit D., and Neve

.1. Neurochem., Vol. 68, No. 3, 1997

1190

A. C. LEBLANC ET AL

R. (1987) Amyloid /3 protein gene: eDNA, mRNA distribution, and genetic linkage near the Alzheimer locus. Science 235, 880—884. Teller J. K., Russo C., DeBusk L. M., Angelini G., Zaccheo D., Dagna-Bricarelli F., Scartezinni P., Bertolini S., Mann D. M. A., Tabaton M., and Gambetti P. (1996) Presence of soluble /3amyloid peptide precedes amyloid plaque formation in Down’s syndrome. Nature Med. 2, 93—95. Thinakaran G., Teplow D., Siman R., Greenberg B., and Sisodia S. (1996) Metabolism of the “Swedish” amyloid precursor protein variant in Neuro2A (N2a) cells. J. Biol. Chem. 271, 9390—

9397. Wasco W., Gurubhagavatula S., Paradis M., Romano D., Sisodia S., Hyman B., Neve R., and Tanzi R. (1993) Isolation and characterization of the human APL2 gene encoding a homo-


logue of the Alzheimer’ s associated amyloid /3 protein precursor. Nature Genet. 5, 95—100. Wertkin A. M., Turner R. S., Pleasure S. J., Golde T. E., Younkin S. G., Trojanowski J. Q., and Lee V. (1993) Human neurons derived from a teratocarcinoma cell line express solely the 695amino acid amyloid precursor protein and produce intracellular /3-amyloid or A4 peptides. Proc. NatI. Acad. Sci. USA 90,9513— 9517. Wisniewski K., Wisniewski H., and Wen G. (1985) Occurrence of neuropathological changes and dementia ofAlzheimer’s disease in Down’s syndrome. Ann. Neurol. 17, 278. Yamazaki T., Selkoe D. J., and Koo E. H. (1995) Trafficking of cell surface /I-amyloid precursor protein: retrograde and transcytotic transport in cultured neurons. J. Cell Biol. 129, 431—442.