Amyloid β Peptide Formation in Cell-free Preparations

Amyloid β peptide (Aβ) is a short peptide that is the major constituent of the amyloid plaques and cerebrovascular amyloid deposits found in Alzheimer's disease. The lack of availability of a cell-free system in which to study Aβ formation has limited our understanding of the molecular mechanisms involved in its production. We report here the reconstitution of such a cell-free system. The reconstituted Aβ formation was temperature-dependent and required ATP. Preincubation with purified protein kinase C (PKC) induced a pronounced inhibition of Aβ formation, similar to that observed in intact cells upon stimulation of PKC. The calmodulin antagonists W-7 and trifluoperazine inhibited Aβ formation and enhanced the action of PKC in both the cell-free system and intact cells. A role for the calcium/calmodulin-activated protein phosphatase calcineurin in the regulation of Aβ formation was demonstrated using a specific peptide inhibitor of calcineurin in vitro as well as cyclosporin A, a cell-permeant inhibitor of calcineurin, in intact cells. Our results suggest that a single substrate might mediate opposing actions of PKC and calcineurin in the regulation of Aβ formation.

Amyloid ␤ peptide (A␤) is a short peptide that is the major constituent of the amyloid plaques and cerebrovascular amyloid deposits found in Alzheimer's disease. The lack of availability of a cell-free system in which to study A␤ formation has limited our understanding of the molecular mechanisms involved in its production. We report here the reconstitution of such a cell-free system. The reconstituted A␤ formation was tempera-

ture-dependent and required ATP. Preincubation with purified protein kinase C (PKC) induced a pronounced inhibition of A␤ formation, similar to that observed in intact cells upon stimulation of PKC. The calmodulin antagonists W-7 and trifluoperazine inhibited A␤ formation and enhanced the action of PKC in both the cell-free system and intact cells. A role for the calcium/calmodulin-activated protein phosphatase calcineurin in the regulation of A␤ formation was demonstrated using a specific peptide inhibitor of calcineurin in vitro as well as cyclosporin A, a cell-permeant inhibitor of calcineurin, in intact cells. Our results suggest that a single substrate might mediate opposing actions of PKC and calcineurin in the regulation of A␤ formation.
One of the salient features of Alzheimer's disease (AD) 1 neuropathology is deposition of the amyloid ␤ peptide (A␤) in brain parenchyma and cerebral vessels. This 40 -42-amino acid peptide is derived from the Alzheimer amyloid protein precursor (APP) (for a review, see Ref. 1). Mutations in APP have been found to cosegregate with affected status in families with earlyonset AD (1). These mutations affect the levels (2)(3)(4), length (5), or primary sequence (6) of the A␤ formed. In all of these cases, it has been argued that the mutations would lead to increased A␤ deposition in the brain in a manner sufficient to cause AD. Thus, an understanding of the mechanisms by which A␤ is formed and the means by which A␤ production is con-trolled may identify avenues for the development of therapies for AD (7).
Numerous studies have demonstrated the formation of A␤ by intact cells (8 -10) and its regulation by a number of signal transduction pathways (reviewed in Ref. 11). Activation of protein kinase C (PKC) and/or inhibition of protein phosphatase 1 or 2A dramatically inhibit A␤ formation (12)(13)(14). Raising intracellular calcium can, according to the experimental conditions, either inhibit or dramatically stimulate A␤ formation (15,16). The mechanisms by which these various signal transduction cascades are able to regulate A␤ formation are currently unknown. The availability of a cell-free system in which to study A␤ formation would greatly facilitate the understanding of the molecular mechanisms involved.
We made use of a Balch homogenizer to prepare cracked cells (17). A␤ production in these broken cell preparations was found to depend on ATP and temperature. By introducing purified PKC, a peptide inhibitor of the calcium/calmodulin-dependent protein phosphatase calcineurin, or a calmodulin antagonist (W-7 or trifluoperazine), it was demonstrated that PKC and calcineurin regulate A␤ formation. A role for calmodulin and calcineurin in the regulation of A␤ formation was confirmed in intact cells by the use of cell-permeant inhibitors. The results suggest that PKC and calcineurin bidirectionally regulate A␤ formation, possibly via regulation of phosphorylation of a single substrate.

EXPERIMENTAL PROCEDURES
Materials-Phorbol 12,13-dibutyrate (PDBu) (from LC Services Corp., prepared as a 1 mM stock solution in dimethyl sulfoxide), W-7 (from Sigma, 10 mM in deionized H 2 O), trifluoperazine (from Calbiochem, 5 mM in deionized H 2 O), cyclosporin A (a generous gift from Drs. G. Snyder and F. Gorelick, 3 mM in dimethyl sulfoxide), calyculin A (from LC Laboratories, 100 M in dimethyl sulfoxide), and calpain inhibitor I (from Calbiochem, 5 mg/ml in ethanol), were diluted to the indicated final concentrations immediately before use. PKC was purified as described (18).
Preparation of Antibodies 3134C and 3129 -Synthetic peptide, corresponding to human A␤1-40, was used as an immunogen to produce precipitating A␤ antibodies in rabbits. One of the antisera (3134) was affinity-purified by making use of a synthetic peptide corresponding to residues 22-41 of human A␤. The affinity-purified antiserum (3134C), which maps to the carboxyl-terminal domain of A␤ (residues 32-40 2 ) and does not recognize secreted APP (sAPP), was used for the ELISA assays. The second antiserum (3129) immunoprecipitated full-length APP and sAPP, as well as A␤. This antibody, which does not cross-react with amyloid precursor-like proteins 1 or 2 because of the absence of A␤ domain (1), was used for immunoblot analysis of sAPP.
Sandwich ELISA for the Detection of A␤-In order to study the regulation of A␤ formation in a cell-free system, a sandwich ELISA was developed. This ELISA makes use of both 3134C as a capture antibody and the monoclonal antibody 6E10 (obtained from Drs. K. S. Kim and H. M. Wisniewski, New York), which is specific for the amino-terminal part of human A␤, as a detection antibody (Fig. 1A). 96-well plates (Nunclon, Maxisorp) were coated for 2 h at 37°C with 10 g/ml 3134C in 10 mM carbonate buffer, pH 8.4. Unoccupied binding sites were then blocked with PBS (1 mM KH 2 PO 4 , 10 mM Na 2 HPO 4 , 137 mM NaCl, 2.7 mM KCl, pH 7.4) containing 1% casein, 0.05% Tween 20, and 0.02% NaN 3 . Plates were washed three times in PBS containing 0.05% Tween 20, 0.02% NaN 3 (PBSTA). Test samples (100 l) were incubated overnight at 4°C. Subsequently, plates were washed three times in PBSTA, incubated for 2 h with 1-3 g/ml 6E10 antibody in PBSTA, washed again, and incubated for 1 h with alkaline phosphatase-linked antimouse-IgG 1 (Southern Biotechnology Associates, Inc.). After three washes in PBSTA, Attophos solution (JBL Scientific, Inc.) was added, and the alkaline phosphatase product was analyzed in a fluorometric plate reader (Millipore Corp.). The ELISA accurately detected Ͼ50 pg/ml A␤, using synthetic peptide corresponding to residues 1-40 of human A␤ as a standard (Fig. 1B).
In some experiments, A␤ was analyzed in conditioned medium from 35 S-labeled cells, using both immunoprecipitation of A␤ and the sandwich ELISA. We compared the levels of A␤ in medium derived from CHO cells expressing wild-type human APP (APP wt ) or a mutated (K670N; M671L; numbering as for APP 770 transcript) form of APP (APP sw ) corresponding to that found in a Swedish kindred (19). An approximately 5-fold increase in A␤ secretion in medium from the cells expressing APP sw , relative to APP wt , was observed with either method (Fig. 1C). Thus, the ELISA accurately detects A␤ in complex biological samples.
Preparation of Cracked Cells-Monolayers of cells expressing APP were washed once in PBS, resuspended in PBS, 5 mM EDTA, centrifuged, washed again in PBS, and resuspended in Buffer H (250 mM sucrose, 15 mM HEPES, pH 7.3, 1 mM magnesium acetate, 1 mM EDTA) to a final concentration of 2 ϫ 10 7 cells/ml. Cells were then passed once through a 0.0004-inch clearance Balch homogenizer (made by H & Y Enterprise) using a tungsten-carbide ball (Industrial Techtonics). Under these conditions, 99.7 Ϯ 0.2% (n ϭ 4) of the cells were trypan blue permeant while still pelleting at 800 ϫ g. These cells, referred to as cracked cells, were incubated (10 7 cells/ml) at the indicated temperatures in the presence of 2.5 mM magnesium acetate, 20 mM KCl, 0.1 mM CaCl 2 , 1 mM dithiothreitol. Energy-depleting (3 mM glucose, 30 units/ml hexokinase) or energy-generating (1 mM ATP, 1 mM GTP, 6 mM phosphocreatine, 80 g/ml phosphocreatine kinase) systems were included in the incubation medium, as indicated. After incubation for the indicated times, cells were chilled on ice and centrifuged at 17,000 ϫ g for 10 min. A buffer solution containing PBS (10 ϫ), 0.5% Tween, 0.02% NaN 3 , 50 g/ml leupeptin, 50 g/ml pepstatin, 20 g/ml trasylol, 1 mM 4-(2-aminoethyl)-benzenesulfonyl fluoride was then added to the supernatant, and A␤ was detected using the sandwich ELISA. Cracked cell preparations were able to produce Ϸ10% of the A␤ secreted by intact cells. In contrast, production of sAPP was almost abolished in cracked cells (Ͻ2% of that produced by intact cells) and was not detectably affected by PKC (data not shown). Thus, in all reported experiments with broken cell preparations, only A␤ formation was studied. Some experiments were carried out with the addition of 100 M GTP␥S (in place of GTP), 2 mM chloroquine, 1 mM N-ethylmaleimide, or 10 g/ml calpain inhibitor I. To study the effect of N-ethylmaleimide, cracked cells were preincubated for 15 min on ice in the presence of 1 mM N-ethylmaleimide, followed by the addition of 2 mM dithiothreitol to quench N-ethylmaleimide. Control preincubations were performed by adding dithiothreitol together with N-ethylmaleimide, as described (20).
Preparation of the reconstituted system-Cracked cells (2 ϫ 10 7 cells/ ml) were centrifuged for 5 min at 1,300 ϫ g at 4°C. After removal of the postnuclear supernatant, the pellet was gently resuspended in ice-cold Buffer H, and incubated for 5 min. This process was repeated three times. After the final resuspension, the cell ghosts were incubated for 15 min at 4°C before constitution with pretreated postnuclear supernatant. The reconstituted system had about 75% of the A␤ forming activity of the cracked cell preparation (not shown).

RESULTS
A␤ Formation in Cell-free Preparations-A␤ formation in cracked cells was dependent on temperature, with little A␤ measured at 4°C compared with the amounts formed at 37°C ( Fig. 2A). A␤ formation was also dependent on the inclusion of ATP and an ATP-regenerating system ( Fig. 2A). Formation of A␤ in cracked cells expressing APP sw or APP wt was linear with time for 30 min, reaching a plateau at about 60 min (not shown). Cracked cells expressing APP sw produced significantly more A␤ than cracked cells expressing APP wt ( Fig. 2A), consistent with results observed in intact cells (Fig. 1C). Since the levels of A␤ formed in cracked cells expressing APP wt were near the limit of detection of the ELISA, we used cells expressing APP sw in additional broken cell studies. The A␤ formed was not precipitated during centrifugation at 100,000 ϫ g for 1 h (not shown), indicating that we measured a soluble form of the peptide. As expected from results obtained in intact cells (21), the formation of A␤ was strongly inhibited by the addition of 10 g/ml calpain inhibitor I (28 Ϯ 6% of control, n ϭ 5, p Ͻ 0.01), demonstrating that A␤ observed in cracked cells is the result of the cleavage of APP occurring during the incubation period. A␤ formation was not affected by substituting a non-hydrolyzable analog of GTP (GTP␥S) for GTP (107 Ϯ 11% of control, n ϭ 4), suggesting that no G-protein was involved in the formation of A␤ in this system. The addition of chloroquine, a weak base that increases the pH of lysosomes and impairs the activity of Inset, the signal obtained for 10 -1000 pg/ml of A␤ (note the semilogarithmic scale). C, cells (20 -30 ϫ 10 5 cells/ml) expressing either APP wt (Wt) or APP sw (Sw) were labeled with [ 35 S]methionine for 2 h and then chased with cold medium for 1 h, as described (12). A␤ in the conditioned medium was either immunoprecipitated with 6E10 (left panel) or detected using the sandwich ELISA (right panel). Immunoprecipitated A␤ was quantified with a PhosphorImager (Molecular Dynamics). Data are expressed as means Ϯ S.E. of four separate experiments. Inset, a typical autoradiogram showing immunoprecipitated A␤. a.u., arbitrary units. lysosomal enzymes (see Ref. 22), did not cause a significant change in A␤ production (94 Ϯ 4% of control, n ϭ 3). The alkylating agent N-ethylmaleimide partially inhibited A␤ formation (71 Ϯ 9% of control, n ϭ 5, p Ͻ 0.05), suggesting the involvement of a free sulfhydryl group in A␤ production. For this reason, 1 mM dithiothreitol was present in the standard incubation buffer of the cracked cells.
To confirm the identity of the protein detected by the sandwich ELISA, we labeled cells expressing APP sw with [ 35 S]methionine. Subsequently, the cells were cracked and incubated under various conditions, and A␤ was allowed to accumulate in the medium. The A␤ from the conditioned medium was then analyzed by immunoprecipitation with 6E10 or 3134C, followed by gel electrophoresis and autoradiography (Fig. 2B). A band with an apparent molecular mass of 4 kDa was present in immunoprecipitates from cracked cells incubated at 37°C in the presence of an ATP-regenerating system. This peptide could be precipitated with either of the two antibodies, and its immunoprecipitation was largely prevented by the addition of excess synthetic A␤1-40 peptide (Fig. 2B). (12), the effect of purified PKC was tested on A␤ formation in the cell-free preparation. Cell ghosts were incubated with postnuclear supernatant that had been preincubated in the presence or absence of PDBu and various concentrations of purified PKC. PDBu alone led to a small but significant inhibitory effect on A␤ formation, presumably due to the activation of endogenous PKC (Fig. 3). Preincubation with PKC in the presence of PDBu dramatically reduced A␤ formation. Inhibition of A␤ formation was also observed when PDBu/PKC was added to cracked cells although this inhibition was somewhat smaller than that observed in the reconstituted system (data not shown). Under the experimental conditions used, cAMP-dependent protein kinase, casein kinase 1, and casein kinase 2 had no effect on A␤ formation using either cracked cells or the reconstituted system (data not shown).

A␤ Formation in Cell-free Preparations Is Regulated by PKC-Since stimulation of PKC by PDBu inhibits A␤ secretion in intact CHO cells
A␤ Formation in Cell-free Preparations Is Regulated by Calcineurin but Not Protein Phosphatase 1 or 2A-Since A␤ for-mation in the cell-free system was regulated by PKC, we investigated the identity of the protein phosphatases involved in this system. Calyculin A (100 nM), a specific inhibitor of protein phosphatases 1 and 2A (IC 50 Ϸ 0.1 nM), inhibited A␤ secretion in intact cells (data not shown). In contrast, calyculin A had no effect on A␤ formation in cracked cells either in the presence or absence of PDBu/PKC (Fig. 4). A third major serine/threonine protein phosphatase, protein phosphatase 2B (calcineurin), is a calcium/calmodulin-activated enzyme. The addition of the calmodulin antagonists W-7 and trifluoperazine, which would be expected to inhibit calcineurin activation, inhibited A␤ formation and augmented the inhibitory effects of PKC on A␤ formation in cell ghosts (Fig. 5). Moreover, preincubation with 50 M calcineurin-inhibitory peptide (23) also inhibited A␤ formation and enhanced the inhibitory effect of PKC (Fig. 4).
A␤ Formation, but Not sAPP Formation, Is Regulated by Calmodulin and Calcineurin in Intact Cells-To assess whether calmodulin and calcineurin also regulate A␤ secretion in intact cells, cells expressing APP wt were incubated in the absence or presence of PDBu, W-7, trifluoperazine, or cyclosporin A (a specific cell-permeant inhibitor of calcineurin) (24). W-7 and trifluoperazine inhibited the secretion of A␤ from intact cells and potentiated the inhibitory effect of PDBu on A␤ formation (Fig. 6). Cyclosporin A had no effect on the basal secretion of A␤ but potentiated the inhibitory effect of PDBu (Fig. 7). On the other hand, calmodulin antagonists (Fig. 6) and cyclosporin A (Fig. 7) had no discernible effect on sAPP secretion in the absence or presence of PDBu. These results indicate that in intact cells PKC regulates both A␤ and sAPP formation, whereas calcineurin regulates only A␤ formation. DISCUSSION One limitation in studying processing and secretion of proteins from intact cells is the difficulty of elucidating the molecular processes involved. In order to better understand the mechanism of action of signal transduction pathways that regulate A␤ secretion, we have developed a cell-free system capable of producing A␤. Since A␤ formation in intact cells requires both cleavage and secretion steps, we made use of a cell-free system that keeps most of the cytoplasmic ultrastructure and the secretory apparatus intact. This procedure, termed "cell cracking," was originally developed to preserve the machinery necessary for studying the release of secretory proteins and prolactin from GH 3 pituitary cells (17). Our results support the conclusion that A␤ is generated by cracked cells, as opposed to residual intact cells for the following reasons: 1) the dependence of A␤ formation on ATP which, when applied extracellularly, has no effect on A␤ formation in intact CHO cells (16); 2) the effect of the cell-impermeant molecules PKC and the calcineurin-inhibitory peptide on A␤ secretion; and 3) the level of A␤ formed (Ϸ10% of that seen in intact cells) compared with the percentage of cells that remained intact (0.3 Ϯ 0.2%).
In agreement with results observed in intact cells, there was much more A␤ produced by cracked cells expressing APP sw as compared with cells expressing APP wt . In fact, it was difficult to detect A␤ formation in cracked cells expressing APP wt , even when using a sensitive sandwich ELISA. For this reason, most broken cell experiments were done with cells expressing APP sw . The question arises as to the relevance of results obtained with these cells for the study of APP wt processing. In experiments in which the two types of transfected cells were compared, they behaved similarly (25). In cell-free preparations, A␤ formation from both APP wt and APP sw was temper-ature-and energy-dependent. Moreover, the involvement of calmodulin and calcineurin in A␤ formation, observed in cracked cells expressing APP sw , was confirmed in intact cells expressing either APP wt or APP sw .
It will be important to establish whether the regulation of APP processing and A␤ secretion established in CHO cells is also observed in cells derived from the brain (e.g. neurons, astrocytes, microglia, and oligodendrocytes). There is a large and conflicting literature concerning the subcellular localization of the machinery responsible for A␤ formation. Hopefully, the development of cell-free preparations, such as the cracked cell preparation in which A␤ secretion is regulated by the same factors that had been found to regulate A␤ secretion in intact cells, will help to resolve this issue. The present results suggest the existence of a phosphoprotein that stimulates A␤ formation. Identification of this protein and determination of its subcellular site of action should help to clarify the locus of A␤ formation.
Stimulation of PKC (12)(13)(14)26) and inhibition of protein phosphatases 1 and 2A (12,14) strongly stimulate sAPP secretion and inhibit A␤ formation in intact cells. It was hypothesized that inhibition of the conversion of APP to A␤ might be attributable at least in part to depletion of substrate resulting from activation of the sAPP metabolic pathway (12,26). In the broken cell preparation used in the present investigation, PKC dramatically inhibited A␤ formation while having no detectable effect on sAPP secretion. These results provide evidence that PKC, in addition to a possible effect on substrate level, has an inhibitory effect on the conversion of APP to A␤. In addition, calcineurin had no effect on sAPP formation but regulated A␤ formation, both in broken cells and in intact cells. Thus, the effects of calcineurin on A␤ formation appear to be independent of substrate depletion resulting from activation of the sAPP Results similar to those shown here were obtained with cells expressing APP sw . B, sAPP formation. sAPP in conditioned medium was analyzed by SDSpolyacrylamide gel electrophoresis and immunoblotting using serum 3129. This experiment is representative of four independent experiments, each carried out in duplicate. metabolic pathway. The loss of regulation by protein phosphatases 1 or 2A of A␤ formation in broken cell preparations is consistent with the substrate depletion hypothesis: the inhibition of protein phosphatases 1 and 2A in intact cells would inhibit A␤ formation solely by increasing sAPP formation. The identification of calcineurin as a major protein phosphatase involved in the regulation of A␤ formation suggests that perturbations of neuronal calcium homeostasis, hypothesized to play an important role in AD (see Ref. 27), might be mediated partially through calcineurin.
Our data are consistent with a model in which PKC and calcineurin, but not protein phosphatase 1 or 2A, can regulate the conversion of APP to A␤ through a mechanism independent of alteration of substrate level. On the other hand, PKC and protein phosphatase 1 or 2A (mainly protein phosphatase 1 (28)), but not calcineurin, regulate the conversion of APP to sAPP. Interestingly, it has recently been shown that a dominant negative mutant of Rab6 affects sAPP, but not A␤, secretion, thus providing independent evidence that sAPP and A␤ are derived from distinct metabolic pathways (29). The increas-ing evidence for such distinct pathways increases the likelihood of developing therapeutic agents capable of inhibiting A␤ formation without affecting sAPP secretion.