Regulated formation of Golgi secretory vesicles containing Alzheimer beta-amyloid precursor protein.

Phorbol esters, activators of protein kinase C (PKC), regulate the relative utilization of alternative processing pathways for the Alzheimer β-amyloid precursor protein (β-APP) in intact cells, increasing the production of nonamyloidogenic soluble β-APP (sβ-APP) and decreasing that of neurotoxic β-amyloid (Aβ) peptide. The molecular and cellular bases of PKC-regulated β-APP cleavage are poorly understood. Here we demonstrate in a reconstituted cell-free system that activation of endogenous PKC increases formation from the trans-Golgi network of secretory vesicles containing β-APP and that this effect can be mimicked by purified PKC. The results demonstrate directly that PKC is involved in regulation of secretory vesicle formation and provide a mechanism by which PKC may reduce the formation of the Aβ peptide characteristic of Alzheimer disease.

molecules of the ␤-APP trafficking and processing apparatus are PKC substrate phosphoproteins involved in the mechanism by which PKC regulates ␤-APP cleavage.
Most ␤-APP resides intracellularly, codistributing with TGN38, a marker of the trans-Golgi network (TGN) (Caporaso et al., 1994). Thus, it seemed possible that PKC might exert its actions on regulated ␤-APP cleavage by redistributing ␤-APP out of its usual residence in the TGN and toward post-TGN compartments where it can undergo processing. This possibility is supported by studies demonstrating stimulation by phorbol esters of the release of glycosaminoglycans, intraluminal molecules of the constitutive secretory pathway (De Matteis et al., 1993;Ohashi and Huttner, 1994). Therefore, we have tested the possibility that an important component of regulated ␤-APP cleavage is PKC-stimulated formation from TGN of constitutive secretory vesicles containing and transporting mature ␤-APP.

MATERIALS AND METHODS
Intact Cell Studies-Three plates of PC12 cells (10-cm dishes, 5 ϫ 10 7 cells/dish) were labeled with 2 mCi/ml [ 35 S]sulfate (Amersham Corp.) for 10 min and chased at 37°C in the presence or absence of 1 M PDBu (LC Services, Woburn, MA) for 5, 15, or 45 min. At the end of the incubation, media were collected and immunoprecipitated with ␤-APP amino-terminal antibody 22C11 (Caporaso et al., 1992). Cells were homogenized, and Golgi-rich membrane fractions were prepared by flotation using a stepwise sucrose gradient (Xu and Shields, 1993). The Golgi-rich fractions were immunoprecipitated with ␤-APP carboxylterminal antibody 369 (Buxbaum et al., 1990). Immunoprecipitates were resuspended in 1% SDS and adjusted to equal amounts of total protein (Bradford), followed by 4 -12% SDS-polyacrylamide gel electrophoresis analysis and autoradiography on Kodak X-Omat TM AR5 film.
Cell-free Vesicle Budding Assay-Confluent PC12 cells (5 ϫ 10 7 ) were pulse-labeled with 2 mCi/ml [ 35 S]sulfate (Amersham Corp.) for 5 min in sulfate-free medium (Life Technologies, Inc.) at 37°C. Cells from each 10-cm dish were homogenized using a stainless steel ball bearing homogenizer (18-m clearance) in 5 volumes of homogenization buffer (0.25 M sucrose, 1 mM magnesium acetate, 0.5 mM EDTA, 0.2 mM CaCl 2 , and proteinase inhibitors) (Xu and Shields, 1993). A postnuclear supernatant was prepared (Tooze and Huttner, 1992) and centrifuged (Beckman TLA-45 rotor) at 14,000 ϫ g for 10 min at 4°C. The pellet was then washed and resuspended in 300 l of homogenization buffer. Aliquots (100 l of membrane preparation) of resulting suspension, the "TGNrich fraction," were incubated in a final volume of 250 l at 37°C for 30 min in the presence of an energy-regenerating system containing 1 mM ATP and 0.2 mM GTP in the absence or presence of 1 mg of cytosol protein/ml (Xu and Shields, 1993). At the end of the incubation period, samples were centrifuged (Beckman TLA-45 rotor) at 14,000 ϫ g for 10 min at 4°C. The pellets and supernatants were separated and immunoprecipitated with antibody 369 and subjected to SDS-polyacrylamide gel electrophoresis and autoradiography. 2 The redistribution of 35 Ssulfated ␤-APP into the supernatant provides a measure of budding of vesicles containing ␤-APP. Treatment of the vesicle-containing supernatant with proteinase K (25 g/ml, Boehringer Mannheim) resulted in disappearance of the ␤-APP-like 22C11 (ectodomain)-immunoreactive material in the presence, but not in the absence, of 1% Triton X-100, consistent with the predicted existence of the ␤-APP ectodomain within a vesicular lumen (data not shown).
Quantification and pairwise analyses were carried out using a Bio-Rad phosphor imaging system (Molecular Analyst TM version 2.0 software). For each experiment, the level of budding observed in the presence of an aliquot of a standard preparation of cytosol was taken as 1 arbitrary unit of budding efficiency, and other levels within a given * This work was supported by The C. V. Starr Foundation (to S. G.), a Cornell Scholar Award in the Biomedical Sciences (to S. G.), and United States Public Health Service Grants AG09464 (to P. G.), AG10491 (to P. G.), and AG11508 (to S. G.). The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
Modulation of Vesicle Budding in the Cell-free Assay-In some experiments, modulation of vesicle budding was tested using (together and separately) purified rat brain PKC (Woodgett and Hunter, 1987) and/or PKC inhibitor peptide-(19 -36) (House and Kemp, 1987). PKC inhibitor peptide-(19 -36), used at 200 M as recommended (House and Kemp, 1987), was synthesized at the Keck Foundation Protein Synthesis and Sequencing Facility at Yale University, New Haven, CT. Effects on vesicle budding were also tested for GTP␥S (30 M) (Boehringer Mannheim) or (AlF 4 ) Ϫ (80 M Al 3ϩ plus 6 mM F Ϫ ).
Stripping of Golgi-rich Membranes Prior to Use in the Cell-free Vesicle Budding Assay-In some experiments, TGN-rich fractions derived from [ 35 S]sulfate-labeled cells were washed with low (control) or high salt and analyzed for vesicle budding in the absence or presence of PDBu (1 M)/purified PKC (25 g/ml) or cytosol (1 mg/ml). For washing of membrane preparations, the [ 35 S]sulfate-labeled TGN-rich fractions were adjusted to 100 or 400 mM potassium acetate and incubated at 4°C for 10 min followed by 10 min of centrifugation (Beckman TLA-45 rotor) at 14,000 ϫ g at 4°C. The pellets were then washed in homogenization buffer, resuspended in buffer of the same composition, and used in the cell-free vesicle budding assay.

Golgi-rich Fractions from Phorbol Ester-treated Cells Are
Depleted of ␤-APP-In intact cells, PDBu caused a depletion of Golgi-associated mature, sulfated ␤-APP (Fig. 1). Sulfated ␤-APP holoprotein content of the Golgi-rich fraction (Xu and Shields, 1993) from metabolically labeled cells chased in the presence of PDBu for 5 or 15 min was approximately 30 and 40%, respectively, of that observed in the absence of PDBu (Fig.  1, left panel). Concomitantly, appearance of s␤-APP in the medium was enhanced severalfold at all times studied (Fig. 1,  right panel).
PKC Activation Stimulates TGN Vesicle Formation in a Cellfree System-Regulation by protein phosphorylation of the biogenesis of ␤-APP-containing vesicles was directly demonstrated using a modification of an in vitro assay of secretory vesicle formation from a TGN-rich fraction (Ohashi and Huttner, 1994). This assay allows study of the effects of cellimpermeant agents such as PKC and GTP␥S. Since it is well established that several cytosolic molecules play important roles in vesicle budding (for review, see Rothman (1994)), budding from isolated TGN was tested in the absence and presence of cytosol.
Cytosol caused a significant increase in vesicle budding from the TGN even in the presence of an optimally effective amount of PDBu/PKC (Fig. 3A, lane 6 versus lane 7), suggesting the presence of a cytosolic factor(s) in addition to PKC. A similar additional stimulatory effect of cytosol was observed when cytosol was added to stripped Golgi membranes (see below) in the presence of optimally effective PDBu/PKC (not shown). The effect of cytosol alone was abolished in the absence of an energy-regenerating system or when incubation was carried out at 20°C, a temperature that blocks TGN exit (Fig. 3B).
GTP␥S and Aluminum Fluoride Inhibit TGN Vesicle Formation in a Cell-free System-Vesicle budding is known to be dynamically regulated (see Bauerfeind and Huttner (1993) for review). Numerous reports have implicated GTP binding proteins (both the small Ras-like class and the heterotrimeric class) in many processes of intracellular vesicular transport including formation of secretory vesicles from the TGN. (AlF 4 ) Ϫ , an activator of both stimulatory and inhibitory trimeric G proteins, diminished the budding of secretory vesicles containing ␤-APP, indicating the involvement of heterotrimeric G proteins. Consistent with this interpretation, GTP␥S, a nonhydrolyzable GTP analogue, also reduced budding (Fig. 3B).
Activated Purified PKC Stimulates Vesicle Budding from Stripped Golgi Membranes, Suggesting That Budding Is Regulated by a Tightly Associated or Integral TGN Phosphoprotein-In order to investigate whether the PKC substrate that regulated budding was likely to be cytosolic or TGN-associated, the [ 35 S]sulfate-labeled, TGN-rich fraction was washed with 100 mM (control) or 400 mM (high salt) potassium acetate. Control and high salt-washed TGN were then assayed for vesicle formation in the absence or presence of purified PKC or cytosol (Fig. 4). The ability of PKC to stimulate vesicle budding in the absence of cytosol was largely retained after stripping of the TGN with high salt. The results suggest that at least one important PKC target is likely to be a tightly associated cytosolic or integral TGN phosphoprotein, presumably acting downstream from various known components of the budding apparatus (Rothman, 1994).
Conclusions-The present results demonstrate that when appropriate PKC-mediated signal transduction systems are activated, ␤-APP is redistributed from the TGN to other cellular locations where it can encounter its processing enzymes, e.g. to the plasma membrane, where it is cleaved by "␣-secretase," another candidate PKC target (Bosenberg et al., 1993;Arribas and Massague, 1995). The data also support the idea that the so-called "constitutive" secretory pathway is subjected to important regulatory influences and that some of this regulation occurs via protein phosphorylation/dephosphorylation (De Matteis et al., 1993;Ohashi and Huttner, 1994). Our results indicate that PKC constitutes an important component of the contribution of cytosol to the regulation of budding. Moreover, the evidence suggests that at least one important PKC substrate is tightly associated with the TGN.
In Alzheimer patients of the Swedish familial AD type, there is an abnormally low ratio of processing of ␤-APP via the non-amyloidogenic s␤-APP pathway relative to the amyloidogenic A␤ pathway (Felsenstein et al., 1994a). This ratio is normalized by activation of PKC, which enhances processing via the non-amyloidogenic pathway while decreasing processing via the amyloidogenic pathway (Felsenstein et al., 1994b;Citron et al., 1994). The present study reveals one cellular mechanism by which PKC produces these effects. In addition, the recent discovery that a major familial AD gene (Sherrington et al., 1995) encodes a protein homologous to the Caenorhabditis elegans sperm molecule spe4, which plays a role in membrane protein sorting (L'Hernault and Arduengo, 1992), raises the possibility that missorting of ␤-APP may contribute to some forms of AD. A more complete understanding of the molecules that control ␤-APP trafficking and processing events, as well as an understanding of how these molecules are regulated, should lead to new insights into the etiology, pathogenesis, and therapy of AD.