The Release of Alzheimer’s Disease p Amyloid Peptide Is Reduced by Phorbol Treatment*

Amyloid precursor protein (APP) is cleaved predomi- nantly within the p amyloid peptide (BAP) domain to release a non-amyloidogenic amino-terminal PN2 frag- ment. Treatment of cells with phorbol dibutyrate, an agent which activates protein kinase C, has been shown to increase the release of an amino-terminal fragment. A panel of mutant APP reporter constructs was expressed in which each of the potential phosphorylation sites lo- cated within the cytoplasmic domain of APP was re-placed with alanine residues. Phorbol response patterns were unchanged for each of these mutants, suggesting that induced cleavage occurs independently of APP sub- strate phosphorylation. We find that phorbol (a) in- creases the release of a PN2 fragment that is consistent with the normal secretase activity, (b) decreases the release of a shorter amino-terminal APP fragment that is cleaved near the amino terminus of BAP, and (c) decreases the release of BAP which was identified based on electrophoretic mobility, epitope mapping, and radio-sequencing. These data demonstrate that pharmaco- logical treatment can reduce the formation of BAP and suggests that protein kinase C activators could be devel- oped as therapeutic agents to block BAP formation. region and the cyto- COOH terminus we use APP-REP to determine ( a ) the nature of the NHz-terminal fragmentb) released from cells after by PDBu, ( b ) whether increased PN2 release is correlated with a reduction in BAP release, and (c) whether We find that the release of PN2 with a corresponding decrease in both the release of a shorter PN2-like NHz-terminal fragment and BAP. These events occur independently of sub-into the extracellular environment and is not intended to imply or infer any other mechanism.

(PN2) into conditioned medium (CM). This cleavage takes place within the BAP sequence (between BAP aa residues 16 and 17) and precludes the proteolytic generation of BAP from APP (10)(11)(12) which is thought to involve an alternative pathway (13). Release of a shorter NHz-terminal APP fragment following cleavage at the amino terminus of BAP (13), a longer NH2terminal fragment following cleavage at a distal site carboxylterminal (COOH-terminal) to the secretase site (14), as well as BAP (15,161, into CM have all been observed. An endosomallysosomal pathway has been suggested to generate potentially amyloidogenic fragments (17)(18)(19). Enhancement of the release of BAP into CM is observed by expressing constructs containing the Swedish KM-to-NL mutation, which flanks the NH2 terminus of BAP (20,21), or 99 aa (Ctss) derived from COOH-terminal APP sequences, which includes BAP and cytoplasmic APP domains (161, Activation of protein kinase C (PKC) is known to regulate the secretion (via proteolytic cleavage) or internalization of a number of membrane proteins (22)(23)(24)(25). Phosphorylation could be involved in regulation of APP processing and the generation of BAP and amyloidogenic fragments, since the APP holoprotein is phosphorylated (26)(27)(28)(29)(30). Treatment of cells with phorbol dibutyrate (PDBu), an agent which activates PKC, increases the release of NHz-terminal APP fragment(s), increases the generation of cell-associated COOH-terminal APP fragments and decreases the amount of mature full-length APP forms, suggesting that substrate APP phosphorylation is involved (26)(27)(28)(29)(30)(31).
To better characterize the mechanisms ofAPP proteolysis, we have developed an APP reporter (APP-REP) as a model system for the expression and cleavage of APP molecules (Ref. 32;Fig. IA). APP-REP is distinguished from endogenously expressed APP by the deletion of 276 central aa of APP and insertion of the Substance P (SP) reporter epitope in the NH2-terminal ectodomain to enable the immunological detection of PN2 fragments released into CM following proteolytic cleavage of substrate. APP-REP contains 113 aa derived from the COOH-terminal portion of APP and includes intact BAP and flanking sequences, the transmembrane spanning region and the cytoplasmic COOH terminus ofAPP. In this paper we use APP-REP to determine ( a ) the nature of the NHz-terminal fragmentb) released from cells after treatment by PDBu, ( b ) whether increased PN2 release is correlated with a reduction in BAP release, and ( c ) whether modulation of PN2 release operates by APP substrate phosphorylation. We find that PDBu treatment specifically increases the release of PN2 with a corresponding decrease in both the release of a shorter PN2-like NHz-terminal fragment and BAP. These events occur independently of subinto the extracellular environment and is not intended to imply or infer any other mechanism. strate phosphorylation within the cytoplasmic domain of APP. Mutations introduced in the NPXY motif (34) of the cytoplasmic domain of APP increases the release of PN2 without reducing the formation of BAP, suggesting that internalization is not required for the formation of BAP and that separate pathways may account for secretase and amyloidogenic activities. These findings suggest that the development of therapeutic strategies to activate PKC in cells generating BAP can reduce the formation of BAP.

EXPERIMENTAL PROCEDURES
Bacterial Strains, Dunsformation, and Plasmid Construction-Transformation, maintenance, and selection of transformants were as described (32). The APP-REP plasmid (pCLL621) represents sequences ofAPP-751 cDNAwith 11  The K"to-NL double mutation (i.e. SEVKMDAEF-to-SEWDAE; APP-REP-NL), which affects the two residuesmmediately N H a~n inal to BAP, and alanine substitutions were introduced into pCLL621 (Figs. 1, A and B) by site-directed mutagenesis using single-stranded phagemid DNA for template as described (33) and resulting mutations confirmed by DNAsequence analysis. C,,,, was constructed in pCLL621 by substituting an NH,-terminal APP-REP fragment, which corresponds to the 5"polylinker (Hind1111 through BAP (EcoRI), with a synthetic HindIII-EcoRI fragment. The fragment encodes APP a a 1-19 (MLPGLALLLLAAWTARALE) fused to MDAEF (sequences near the NH,-terminal region of BAP) and provides the native signal peptide and cleavage site and reconstructs native BAP sequence. Following cleavage of the signal peptide, a 102-aa protein encoding LE and 100 a a derived from the COOH terminus of APP is expressed.
Tissue Culture Lines and Dansfection-All cells including SV40transformed monkey kidney COS-1 cells (CRL 1650) for transient expression, and human embryonic kidney 293 (CRL 1573) and glioblastomdastrocytoma U-87 MG (HTB 14) cells for stable expression systems, were obtained from American w e Culture Collection and maintained according to their recommendation. Transfection and selection conditions were as described (32).
Radiolabeling of Cells and Preparation of Cell Lysates and Conditioned Medium-APP-REP proteins expressed in exponentially growing monolayers of adherent cells were radiolabeled for the times indicated by the metabolic incorporation of 0.15-0.5 mCi of [35Slmethionine (800 Ci/mmol; Amersham Corp.) a s described (32). AMe2S0 solution with or without phorbol dibutyrate (PDBu; Sigma) was added to labeling or chase medium as indicated (final concentrations: 0.05% MezSO with or without 1 p PDBu). For labeling of cells in suspension, cell monolayers were washed twice with 4 ml of labeling medium (methionine-free medium supplemented with 2% dialyzed fetal bovine serum and 25 m HEPES, pH 7.4) and incubated for 30 min a t 37 "C to deplete methionine. Cells were suspended by gentle trituration, pelleted, and resuspended in 2 ml of labeling medium and incubated a t 37 "C with 0.15-0.5 mCi of [35Slmethionine for the times indicated. For pulse-chase experiments, a n excess of ice-cold chase medium (labeling medium supplemented with 1 m unlabeled methionine) was then added and the cells washed twice by centrifugation at 4 "C. Labeled cells were then resuspended at 4 "C in fresh chase medium supplemented with a MezSO solution withlwithout PDBu as before and incubated at 37 "C for the times indicated. Supernatants were collected and CM and cell lysates prepared (-4 x IO6 celldl0-cm culture disW5 ml of CM or lysate) as described (32).
For radiosequence analysis, two 10-cm culture dishes of 293 cells stably overexpressing the indicated constructs were double-labeled for 6 h by the incorporation of 0.5 mCi of [36Slmethionine and 0.5 mCi of [3Hlphenylalanine (Amersham Corp.), CM-immunoprecipitated with antibody 6E10, and immune complexes fractionated by SDS-polyacrylamide gel electrophoresis (&20% Tris-Tricine gels; Daiichi, Tokyo, Japan) as usual (32). Proteins were then transferred to membrane, visualized by autoradiography, and the -3.5 kDa bands excised and subjected to radiosequence analysis essentially as described (15,16).
Antisera a n d Immunoprecipitation Analysis-Rabbit polyclonal antisera to the SP reporter epitope, immunoprecipitation of radiolabeled lysate or CM, was as described (32). Anti-mouse IgG,-agarose (Sigma) was used for the precipitation of monoclonal antibodies 6E10 (anti-BAP1"16; Ref. 35) and 4G8 (anti-BAP,,_,,; Ref. 36 (33). Codons are identified by numbers according to APP-751 and represent sequences corresponding exactly to the cytoplasmic domain of APP. The alanine substitutions generated are referred to a s Y709A, T710A, S711A, T724A, S731A, Y738A, T742A, and Y743A. The underlined motif represents the "NPXY" sequence putatively analogous to the internalization consensus sequence of low density lipoprotein receptor (34).
plexes, autoradiography, and quantitation by scanning laser densitometry or phosphorimage analysis (Molecular Dynamics, Sunnyvale, CAI was as described (32).  Fig. 4) we find that the effect of PDBu on the progression of post-translational modifications, cleavage of cell-associated full-length APP-REP, and NH2-terminal APP-REP fragment release are nearly identical (data not shown) to those described for APP counterparts (26)(27)(28)(29)(30)(31). The presence of 1 staurosporine, an inhibitor of PKC, also eliminates the PDBu response with APP-REP (data not shown). In summary, activation of PKC increases the fraction of   (Fig. LA) to characterize the type of NH2-terminal APP fragment(s) released by treatment with PDBu (Fig. 3). Ordinarily only an -67-kDa band is visualized (Fig. 3A, lams 2 and 31, but closer examination of lightly exposed autoradiograms reveals the presence of a doublet band migrating at -65-67 kDa. Expression of an APP-REP derivative, which is cleaved poorly by secretase (substitution of BAP1,-z, with 11 aa derived from another reporter e p i t~p e ) ,~ reveals a resolved doublet and a shift in abundance toward the faster migrating component (i.e. -65 kDa; Fig. 3 A , lanes 4 and 5). We then tested the APP-REP fragments released into CM for the presence of the NH2-terminal portion of BAP ( i e . BAP aa residues 1-16) by differential immunoprecipitation with 6E10 (anti-BAP1-16) antibody (Fig. 3B). Immunoprecipitation of CM from untreated control cells with 6E10 yields predominantly the upper component of the doublet (lane 4 ) as compared with precipitation with SP (lane 3 ) . Immunodepletion of CM with 6E10 (lane 4 ) and subsequent immunoprecipitation of the remaining CM with SP (lane 5 ) clearly reveals the lower faster migrating -65-kDa band. These data are consistent with the report that secretory processing of APP results in the release of classical PN2 containing BAP1-16 as well as a shorter derivative whose complementary cell-associated fragment contains an intact BAP sequence (13).   (lanes 4 and 5 ) , or vector only control (lane 1) was immunoprecipitated with SP. An autoradiogram representing a short exposure is given to demonstrate the appearance of a doublet band (lanes 2 and 3 ) which, upon longer exposure, becomes unresolved and is ordinarily observed as a singlet band (e.g. Fig. 2, A and B, lanes 5 and 6 )  we fail to detect the faster migrating -65-kDa band. This indicates that PDBu preferentially enhances the release of fulllength PN2.

Deatment of Cells with PDBu Increases the Release
PDBu-enhanced Release of PN2 Is Correlated with Reduced Formation of BAP-To detect and validate the identity of BAP, we analyzed a larger volume of CM either from 293 cells stably expressing APP-695 and APP-REP (Fig. 4) or COS-1 cells transiently expressing the same constructs (data not shown). Immunoprecipitation of CM with 6E 10 antibody (Fig. 4A ) reveals the presence of -4.5 (lanes 5 and 7) and -3.5 (lanes I-10)-kDa fragments. The -4.5-kDa fragment is found only in the CM of cells transfected with APP-REP and is uncharacterized. The -3.5-kDa fragment is detected in CM of all cells but is more abundant in CM of cells transfected with APP-695 (lanes 1 and 2 ) , APP-REP (lanes 5 and 61, or the mutant APP-REP derivative Y743A (lanes 7 and 8) described below. In the presence of competing cold synthetic BAPlao, the failure to precipitate either fragment with 6E10 antibody indicates both contain an epitope of BAP (data not shown). Only the -3.5-kDa fragment (and a shorter -3-kDa fragment; Ref. 11) is precipitated with 4G8 antibody (data not shown). Therefore, epitope mapping with antibodies having known specificity for BAP sequences and the detection of an -3.5-kDa fragment in CM from cells overproducing APP-REP and APP-695, an observation similar to that reported by others (16, 37, 3 8 , provides supporting evidence that our -3.5-kDa peptide is BAP. Radiosequence analysis of -3.5-kDa fragments derived from a panel of 293 cells stably overexpressing APP-695, APP-REP, APP-REP-NL, and Y743Aconfirms the identification of the -3.5-kDa band as BAP (Fig. 4 0 (lanes 1.3, 6, 8, and 9 ) or control treated (lanes 2, 4, 5 To determine the effect of PDBu upon formation of BAP, 293 cells were treated with PDBu and CM compared with untreated controls for the presence of immunoprecipitable BAP (Fig. 4A, compare lanes 2 and 1,5 and 6, and 7 and 8). PDBu reduces the amount of -3.5-kDa fragment by 60-70% (Fig. 4B) and -4.5-kDa fragment by 80-90% (data not shown). These data demonstrate that 293 cells normally release BAP into CM and treatment with PDBu causes a significant reduction in release of immunoprecipitable BAP (-3.5-kDa fragment). We observe similar patterns of BAP expression with COS-1 cells (data not shown).

FIG. 4. Production and release of BAP by 293 cells into CM and effect of PDBu treatment on BAP formation. A, immunoprecipitation analysis of CM from PDBu
Construction and Expression of APP-REP-based "Phosphorylation-minus" APP Mutants-If phosphorylation of APP is the event which alters processing, mutations introduced at critical sites to prevent phosphorylation should block the observed PDBu response. To construct such phosphorylation-minus derivatives, we substituted each of the 8 aa that are po-tential phosphorylation substrates located within the cytoplasmic domain of APP-REP with alanine (Fig. 1B). The panel of mutant derivatives were then stably expressed in HTB14 cells. With the exception of Y743A (see below), each mutant releases basal levels of PN2 similar to that of wild type APP-REP, and all typically display a 3-4-fold increase in release of PN2 in response to PDBu (Fig. 5). Quantitation of cell-associated full-length forms indicates similar levels of expression for each mutant construct (data not shown). We observe an identical pattern of PDBu response with wild type APP-REP and the mutant derivatives expressed stably in 293 or transiently in COS-1 cells (data not shown). The fact that phosphorylation-minus mutations continue to be responsive to PDBu indicates that phosphorylation of the APP cytoplasmic domain by PKC (or tyrosine kinase) is not required for PDBustimulated release of PN2.
Expression levels of cell-associated full-length Y743A (Fig.  1 B ) are similar to wild type APP-REP (data not shown). However, the release ofY743A-derived PN2 is -3-4-fold more than untreated wild type APP-REP controls, whereas addition of PDBu results in only a minimally enhanced release of PN2 Release of p Amyloid Peptide Is Reduced by Phorbol Deatment ectodomain cleavage and secretion independent of substrate phosphorylation following the activation of PKC by treatment with phorbol ester (40)(41)(42). If phosphorylation directly activates secretase, labeling with 32P or binding with PKC (43) may provide an avenue for the characterization, isolation, and cloning of secretase. Alternatively, phosphorylation of an unidentified target may regulate APP processing by increasing access to secretase in a manner which competes with NH2-/COOH-terminal BAP cleaving activities for APP substrate.
The substituted tyrosine of Y743A, which influences APP processing, is located within NPXY, a motif that may be a homolog to the cytoplasmic sequence on the low density lipoprotein receptor which mediates internalization of receptors from the cell surface by coated pit formation (34). This is consistent with the recent observations demonstrating that APP is internalized (17) and localized in clathrin-coated vesicles (44). Deletion of the entire cytoplasmic domain of APP (involving 3 tyrosine residues; Ref. 201, or a portion of cytoplasmic APP sequence (YENPTY; Ref. 451, prevents the internalization of APP and increases the release of PN2. Although these studies imply that NPTY functions similarly to the NPXY motif of low density lipoprotein receptor, the expression of Y743A with a single tyrosine substitution better demonstrates the involvement of the second tyrosine in signaling internalization rather than the first ( i e . YENPTY) as has been proposed (46). These studies support the-view that cleavage ofAPP by secretase may occur on the plasma membrane; however, it has not been demonstrated that failure to internalize results in an increase of APP on the plasma membrane and release of PN2 from that substrate. Furthermore, certain studies suggest that cleavage ofAPP by secretase may occur within an intracellular compartment (46)(47)(48). Alternatively, loss of NPTY function could redirect the transport of more APP to the cell surface by default rather than into the endosome-lysosome pathway (49). In this case, more APP substrate entering the secretase pathway can be cleaved either on the plasma membrane or intracellularly. It is likely that the APP cytoplasmic domain participates in multiple roles pertaining to APP trafficking and processing.
Our data suggest that the inverse relationship between release of PN2 and BAP observed with PDBu treatment is complex as exemplified by the comparison of APP-REP with its mutant derivative Y743A and control or PDBu treatment. Although the same PDBu-induced release of PN2 from APP-REP is not observed with Y743A, we speculate that the already robust release of PN2 with untreated Y743A masks the potential for a further PDBu response due to either saturation of secretase or limited availability of additional substrate. However, elevated release of PN2 with untreated Y743A was not associated with a concomitant decrease of BAP release compared with the untreated APP-REP control. We interpret this lack of a n inverse correlation as follows. First, our observations suggest that generation of BAP may not require internalization. Second, cleavage of APP by secretase occurs either subsequent to, or independent of, the generation of BAP. A model consistent with our data is graphically depicted in Fig. 6 to demonstrate these points. ( Fig. 5). Furthermore, measurement of BAP released into CM indicates that Y743A produces similar levels of BAP compared with APP-695 and APP-REP and that release is down-regulated by PDBu to a similar extent (Fig. 4, A and B). A greater amount of the -4.5-kDa fragment is observed with Y743A ( Fig.  4 A , compare lanes 7 and 5 ) . DISCUSSION Current understanding of the mechanisms of proteolytic processing of APP suggest that two different pathways may account for the release of ( a ) PN2, following cleavage by secretase (Le. "secretase pathway"; Refs. 10 and ll), and ( b ) BAP, following dual cleavage by activities acting a t both NH2-and COOH-terminal sequences of BAP ( i e . "alternative pathway"; Ref. 13). The data presented here add to an understanding of these two processes. The activation of PKC by treatment with phorbol esters enhances the release of NH2-terminal APP fragment(s) which was implied (30,31,39) and now shown (this work) to represent PN2 and not a shorter potentially amyloidogenic PN2-like derivative (13).
Our demonstration that PDBu, while increasing the release of PN2, simultaneously decreases the release of BAP is significant. The obvious interpretation of our data is that the secretase and alternative pathways compete for APP substrate. However, expression of the mutant APP derivative Y743A and other data requires consideration of other models.
We have clearly demonstrated that PDBu, a known activator of PKC, increases the release of PN2 by a mechanism which is independent of cytoplasmic phosphorylation of APP. Therefore, it is likely that phosphorylation of secretase or other proteins regulates the processing of APP and perhaps other membrane spanning receptors. It is interesting to note that a number of membrane-spanning receptors, such as pro-transforming growth factor-a and the CSF-1 receptor, also undergo rapid ( Fig. 6, A-D). This model allows for the possibility that secretase cleaves a pool of APP which is independent of the one which participates in the generation of BAP or that all APP may pass through a single "secretory pathway" consisting of early and late components which define the "alternative" and "secretase" pathways. In one variation, phosphorylation of an unknown target may reduce the exposure of APP to NH2-/ COOH-terminal BAP cleaving activities (alternative pathway) by rapid redistribution ofAPP substrate into the next compartment (secretase pathway). It is known that phosphorylation alters trafficking. For example, phosphorylation apparently enhances the availability of synaptic vesicles for neurotransmitter release (50) and alters receptor traffic and pool size by facilitating the redistribution of certain receptors to plasma membrane and others to intracellular membranes in response to PKC activation (51, 52). The -3-kDa fragment derived from the COOH-terminal portion of BAP, possibly the result of secretase and COOH-terminal BAP cleaving activities (15,38), may in fact represent a fragment derived through a single secretory pathway for BAP and PN2.
Aside from the mechanistic issues, the decrease in release of BAP by PDBu demonstrates that BAP formation can be pharmacologically reduced and suggests a drug discovery strategy for developing therapeutics using tissue culture models. We predict that receptor-mediated activation of PKC will also elicit the same response. Cells expressing muscarinic acetylcholine m l or m3 receptors increase the release of NH2-terminal APP fragment(s) in response to the cholinergic agonist carbachol (39,53). Increased release is blocked either by the muscarinic antagonist atropine (39,53) or the PKC inhibitor staurosporine (53), but not by calcium ionophoreA.23187 (53). By co-expressing APP-REP and rat m l receptor, we have confirmed t h a t m l agonists increase release of PN23 and are testing for the inverse relationship down-regulating the formation and release of BAP. Similarly, interleukin-1, a cytokine that is thought to mediate APP expression via PKC (54), activates a receptor-PKC coupled increase in APP release (39). These observations indicate that receptor-mediated PKC activation, or regulation of the targets of phosphorylation, might be exploited for developing therapeutic interventions that prevent the formation or release of BAP. It has been reported recently, after completion of this work, that protein phosphorylation inhibits the production of BAP. In that report, compounds known to activate phospholipase C, and therefore PKC, were shown to reduce the generation of BAP in a model which is similar to and supportive of our own (55). However, the actual mechanism responsible for the PDBu response and the apparent inverse relationship between formation of PN2 and BAP is as yet unknown.