Processing of the @-Amyloid Precursor - The Journal of Biological ...

THE JOURNAL OF BIOLOGICAL CHEMISTRY Q 1993 by The Arnericnn Soeiety for Bioehemistry and Molecular Biology, Inc.

Vol. 268, No. 22, Issue of August 5, pp. 16602-16609, 1993 Printed in U.S. A.

Processing of the @-AmyloidPrecursor MULTIPLEPROTEASESGENERATE

AND DEGRADE POTENTIALLY AMYLOIDOGENIC FRAGMENTS* (Received for publication, December 21,

1992, and in revised form, April 13, 1993)

Robert SimanS, Suzanne Mistretta, John T. Durkin, MaryJ. Savage, Tatjana Loh, Stephen Trusko, and RichardW. Scott From Cephalon, Inc., West Chester, Pennsylvania 19380 Proteolytic processing of the @-amyloidprecursor proteins (APP) is required for releaseof the @/A4protein and its deposition into the amyloid plaques characteristic of aging and Alzheimer’s disease. We have examined the involvement of acidic intracellular compartments inAPP processing in culturedhuman cells. The use of acidotropic agents and inhibitors toa specific class of lysosomal protease, coupled with metabolic labeling and immunoprecipitation, revealed that APP is degraded within an acidic compartment toproduce at least 12 COOH-terminal fragments. Nine likely contain the entire @/A4domain and, therefore, are potentially amyloidogenic. Treatment with E64 or ZPhe-Ala-CHNz irreversibly blocked activities of the lysosomal cysteine proteases cathepsins B and L but did not inhibit thelysosomal aspartic protease cathepsin D and did not alter the production of potentially amyloidogenic fragments. Instead, the inhibitors prevented further degradation of the fragments. Thus, large numbers of potentially amyloidogenic fragments of A P P are routinely generated in an acidic compartment by noncysteine proteases and then are eliminated within lysosomes by cysteine proteases. Immunoblot and immunohistochemical analyses confirmed that chronic cysteine protease inhibition leadsto accumulation of potentially amyloidogenic APP fragments in lysosomes. The results provide further support for the hypothesis that anacidic compartment may be involved in amyloid formation and begin to define the proteolytic events that may be important for amyloidogenesis.

(Kang et al., 1987; Ponte et al., 1988, Tanzi et al., 1988 Kitaguchi et al., 1988). Synthetic peptides corresponding to regions of @/A4or the @/A4protein itself spontaneously form aggregates in solution that, ultrastructurally, resemble the fibrils comprising amyloid deposits of Alzheimer’s disease (Castano et al., 1986; Kirschner et al., 1987). Since @/A4 protein alone can form an amyloid fibril, attention hasfocused on mechanisms of the genesis of @/A4 protein. It is now established that the @/A4protein is not directly synthesized by cells, butratherthe proteolytic processing of APP is necessary for @/A4formation. Thus, an understanding of the cellular trafficking and processing of APP is essential for elucidating amyloidogenesis. APP is aconstituent of many cell types, anditsposttranslational processing has been examined in a variety of cultured cells. All of the DIA4-containing APP isoforms possess a single membrane-spanning domain and a signal sequence at theextreme NH2 terminus,which directs co-translational insertion intothe endoplasmic reticulum (Kang et al., 1987; Dyrks et al., 1988). Newly synthesized APP is delivered to the Golgi, where tyrosine sulfation and 0-linked glycosylation occur (Schubert et al., 1989; Weidemann et al., 1989; Oltersdorf et al., 1990; Caporaso et al., 1992). After this stage, APP processing diverges into distinctpathways. In one route, a large NH2-terminal APP fragment is generated by cleavage within the @/A4domain, between residues Lys-16 and Leu17 (Esch et al., 1990; Anderson et al., 1991), and is constitutively secreted (Weidemann et al., 1989). Since this secretory route destroys the @/A4domain, it may be a non-amyloidogenic processing pathway. Pharmacological, histochemical, and subcellular fractionation experiments indicate that APP is also processed within a second pathway, in the endosomal-lysosomal system (Cole Protein deposition occurs in the extracellular space and et al., 1989; Golde et al., 1992; Caporaso et al., 1992; Haass et vasculature of the brain as a normal consequence of aging and al., 1992a). Inhibition of APP processing within this pathway is markedly increased in Alzheimer’s disease and Down’s causes accumulation of COOH-terminal fragments of approxsyndrome (Selkoe, 1991). These deposits, referred to as amy- imately 9,9.5,11.5,12 (Golde et al., 1992), and 22 kDa (Haass loid, are composed in large part of a 39-42-amino acid protein et al., 1992a). Several observations point to the potential known as @/A4 (Glenner and Wong, 1984; Masters et al., importance of lysosomes, and of APP processing within them, 1985), which is a small domain contained within a family of in the pathogenesis of @/A4protein. First, the larger APP larger proteins, the @-amyloid precursor proteins (APP)’ derivatives found within lysosomes contain the entire @/A4 domain and thus are potentially amyloidogenic (Estus et al., * The costs of publication of this article were defrayed in part by 1992). Second, the concentration of a -19-kDa COOH fragthe payment of page charges. This article must therefore be hereby ment is elevated with human aging (Nordstedt et al., 1991); marked “aduertisernent” in accordance with 18 U.S.C. Section 1734 this APP derivative may correspond to one of the COOHsolely to indicate this fact. fragments found within endosomes-lysosomes. $ To whom correspondence should be addressed Cephalon, Inc., terminal 145 Brandywine Pkwy., West Chester, PA 19380.T e l 215-344-0200; Third, COOH-terminal APP fragments similar in size to those Fax: 215-344-0065. found within endosomes-lysosomes are reportedly neurotoxic The abbreviations used are: APP, ,%amyloid precursor protein; (Yankner et al., 1989; Neve et al., 1992) and, when overexMEM, modified Eagle’s medium; PAGE, polyacrylamide gel electrophoresis; Tricine, N-[2-hydroxy-l,l-bis(hydroxymethyl)ethyl]gly- pressed in cultured cells, lead to amyloid fibril formation (Maruyama et al., 1990). Fourth, aggregation of @/A4protein cine; E64, transs-epoxysuccinyl-~-leucylamido(4-guanidino)-butane; is promoted at the acidic pH level of the lysosome (Burdick Z, benzyloxycarbonyl; AMC, aminomethylcoumarin.

16602

APP Degradation in Acidic Compartments et al., 1992).Last, anumber of lysosomal hydrolases massively accumulate within dystrophic neurons in Alzheimer's disease brain and are closely associated with amyloid deposits (Cataldo et ul., 1990,1991).Collectively, the results have prompted the hypothesis that altered lysosomal processing of APP may lead to amyloid formation and deposition (Benowitz et al., 1989; Cole et al., 1989; Cataldo et al., 1990; Goldeet al., 1992; Haass et al., 1992a). In viewof the evidence linking the endosomal-lysosomal system t o the pathogenesis of amyloid deposits, it is important to define specific proteolytic events within these and other acidic intracellular compartments that regulate the content of potentially amyloidogenic fragments of APP. Among the large number of lysosomal hydrolases are four proteinases: the aspartic protease cathepsin D, and the cysteine proteases cathepsins B, L, and S (Barrett and Kirschke, 1981; Wiederanders et al., 1992). Here, we report on the effects of acidotropic agents that block lysosome function and inhibitors of lysosomal cysteine proteases on the production and clearance of APP fragments in cultured human cells. The use of irreversible inhibitors permitted simultaneous evaluation of the blockade of specificlysosomal proteases and alteration in APP degradation. We have found that atleast nine potentially amyloidogenic APP fragments are routinely generated and then rapidly degraded within acidic compartments. Moreover, there is a segregation of protease function; noncysteine protease(s) degrade intact APP toproduce the potentially amyloidogenic fragments, while the family of lysosomal cysteine proteases serves to further degrade them. MATERIALS ANDMETHODS

Stably Transfected Cells Overexpressing APP751-A cDNA encoding full-length human APP751 (Ponte etal., 1988; Tanzi et al., 1988; Kitaguchi et al., 1988) wascloned from a human brainlibrary (Savage et al., 1993).The cDNA was inserted into a plasmid expression vector containing a cytomegalovirus promoter and neomycin resistance selectable marker, and the construct was used to transfect the human kidney 293 cell line by CaPO, co-precipitation (Savage et al., 1993). Cells were maintained in MEM/10% horse serum. After 72 h, transfectant selection was begun by addition of G418 at 1 mg/ml. At 14 days post-transfection, individual clones were picked and subcloned by limiting dilution. APP expression and secretion were evaluated by immunoblot analysis (see below) of cell lysates and conditioned media from 60 clones. A recombinant APP segment corresponding to the COOH-terminal 100 residues was prepared from a bacterial expression system as previously described (Savage et al., 1993). Antibodies-The 22Cll monoclonal antibody recognizes an APP epitope near the NH2 terminus (Boehringer Mannheim). Antibodies 10E and 12E were raised to synthetic peptides corresponding to domains near the APP COOH terminus. The peptides (APP705-728 for AblOEAPP722-751 for AblSE, numbering according to APP751) were synthesized by solid phase methods, and their structures were confirmed by fast atom bombardment-massspectrometry. For use as immunogens, the peptides were conjugated to bovine serum albumin with glutaraldehyde. Antibodies were raised in rabbits and affnitypurified from immune sera by conjugating the peptides to thyroglobulin and coupling the conjugates to CNBr-activated Sepharose (Pharmacia LKB Biotechnology Inc.). Immunoblotting and Immunoprecipitation-Tris/Tricine SDSPAGE was used for improved resolution of APP fragments of 3-30 kDa (Schagger and von Jagow, 1987; Estus et al., 1992). Cells were harvested in 20 mM Tris-HC1 (pH 7.4), 1mM EDTA, 50 p~ leupeptin, 1 p M pepstatin A for blotting experiments. Immunodetection of APP and its fragments was carried out following electrophoretic transfer to polyvinylidene difluoride membranes (Bio-Rad), as previously described (Siman et al., 1990). 22Cll was used a t 1:300, while 10E and 12E were diluted 1:500.For comparison of the efficiencies of APP secretion among clones of stable transfectants, serum-free media were conditioned for 1 h and concentrated by acid precipitation prior to SDS-PAGE. APP was metabolically labeled and immunoprecipitated using

16603

standard methods (Weidemann et al., 1989; Oltersdorf et al., 199% Golde et al., 1992; Caporaso et al., 1992). Briefly, culture dishes (10 cm) were washed and incubated for 10 min in methionine-free MEM (Life Technologies, Inc.) andthen were incubated in 2.5 ml of methionine-free MEM containing 250 pCi of [36S]Met(Du Pont) for periods ranging from 30 min to 3 h. For chase experiments, cells were washed twice with Hepes-buffered Hanks' balanced salt solution and incubated innormal serum-containing medium until harvest. In other experiments, APP and its fragments were directly immunoprecipitated without prior labeling. Cells were extracted in 50 rnM Tris-HC1 (pH 8.0), 150 IIIM NaCI, 1% TritonX-100,0.25% deoxycholate, 0.1% SDS, and the protease inhibitors leupeptin (50 pM), pepstatin (20 (2 mM), benzamidine (1mM), and phenylmethylsulfonyl p ~ )EDTA , fluoride (0.25 mM). Each extract was incubated with 5 pl of 10E or 12E, followed by 10 pl of protein A-Staphylococcusaureus (Pansorbin, Calbiochem). For preabsorption of 12E prior to immunoprecipitation, 5 pl of the antibody was incubated for 30 min at 22 "C with 100 r g of peptide APP722-751, and then themixture was added to cell extract. Following electrophoresis, gels were subjected to fluorographic enhancement, dried, and used to expose x-ray film. Individual polypeptides were quantified by densitometry. Immunohistockemistry-293A14 cells were grown on glass slides coated with polylysine/laminin. Some cultures were treated for 1.5 days with E64 a t 100 p~ prior to immunostaining. Cells were fixed for 10 min in ice-cold 4% paraformaldehyde, 0.1 M phosphate (pH 7.4) and permeabilized for 10 min with 0.1% Triton X-100, 50 mM Tris-HC1 (pH 7.4), 150 mM NaCl, 5% horse serum. APP COOHterminal epitopes were labeled by a biotin-avidin-horseradish peroxidase procedure (Simanet aL, 1990), employing rabbit Abl2E as primary antibody (1:500). Late endosomes and lysosomeswere stained with an antibody to human cathepsin D (1:1000, Calbiochem). Quantitation of Protease Actiuities"293A14 cells were treated for 2 h with various concentrations of Z-Phe-Ala-CHN2 and then were washed three times in Hepes-buffered Hanks' balanced salt solution, collected by scraping into assay buffers, and sonicated. Cathepsin B activity was measured using Z-Arg-Arg-AMC at 100 p~ as substrate, while cathepsin L activity was measured using Z-Phe-Arg-AMC at 100 FM as substrate (Barrett and Kirschke, 1981). In both cases, substrate hydrolysis was quantified by spectrofluorimetry under conditions in which lysis increased linearly as a function of incubation time. Assays conducted with purified cathepsins B and L confirmed that the reaction conditions were selective for the two proteases. Cathepsin D activity was measured in 0.4 M sodium acetate (pH 3.2) using ["C]methemoglobin as substrate (Takahashi and Tang, 1981). Activity was defined as the difference in acid-soluble radioactivity between reactions run in the absence and presence of the aspartic protease inhibitor pepstatin A (5 p ~ ) . RESULTS

StablyTransfected Human Cell Lines That Differentially expression construct coding forhuman APP751 and neomycin resistance was transfected into the human kidneycell line 293, and stable transfectants were selected and subcloned. Sixty cloneswere examined for APP751 expression and secretion by immunoblotting of cell lysates and conditioned media with the monoclonal antibody 22Cl1, which labels both full-length and truncated, secreted APP (Weidemann et al., 1989). Considerable variation was observed not only in the level of APP751 overexpression but also in the relative levels of cell-associated uersus secreted APP and in the forms of APP. Fig. 1 illustrates the disparity in APP processing between two of the clones. APP751 was efficiently cleaved and secreted from the 293A10 clone, as evidenced by the large amount of APP detected in the medium (-120 kDa) and thelack of detectable full-length fully mature APP (-130 kDa) in the cells. Immunoblotting with an antibody directed at a COOH-terminal domain confirmed that the secreted APP was truncated at its COOH terminus (data not shown). By comparison, the 293A14 clone secreted approximately 10-fold less APP, despite the nearly equivalent levels of newly synthesized cell-associated APP (-110 kDa). Decreased APP secretion was accompanied by an accumulation of the full-length mature -130-kDa APP in cell lysates. Process APP751-An

APP Degradation in Acidic Compartments

16604

to their effects on the 9-kDa fragment, chloroquine and especially ammonium chloride greatly reduced the content of the five larger fragments. This is fully consistent with previous Mr ~ 0-3 l observations that APP cleavage within a secretory pathway generates a single 9-kDa COOH-terminal fragment, while additional larger fragments are produced within an acidic compartment such as endosomes or lysosomes (Caporaso et al., 1992). Processing of APP by Lysosomal Cysteine Proteases-To define the role of specific classes of lysosomal protease in 106 APP degradation, 293A14 cells were treated with the irreversible inhibitors of lysosomal cysteine proteases E64 and Z80 Phe-Ala-CHNz.The former is highly selectivefor all cysteine proteases (Barrett et al., 1982), while the latter is a potent and selective blocker of the lysosomal cysteine proteases cathepsins B and L (Kirschke and Shaw, 1981; Green and 47 Shaw, 1981). As shown in Fig. 2, both inhibitors caused an increase within 3 h in several labeled APP COOH-terminal derivatives, including 9-, lo-, 12-, 13-, 15.5-, 19-, and 22-kDa polypeptides. Maximal effects were observed with 100PM E64 and 1PM Z-Phe-Ala-CHNz (data notshown). 293A10 To examine whether E64 and Z-Phe-Ala-CHNz were affecting APP processing by inhibiting endosomal-lysosomal proteases, the inhibitors were combined with the acidotropic 293A14 FIG.1. Overexpression of APP761 by stably transfected agents. Neither E64 nor Z-Phe-Ala-CHNzcaused increases in human cells. 293 cells overexpressing human APP751 were gener- APP fragments in the presence of either ammonium chloride ated as described under “Materials and Methods.” Shown here are or chloroquine (Fig.2). Therefore, the lo-, 12-, 13-, 15.5-, cell lysates and conditioned media from two of the clones Western 19-, and 22-kDa APP COOH-terminal derivatives likely are blotted and stained with 22Cl1, an antibody that recognizes both full-length andCOOH-terminal truncated APP. Molecular mass generated within an acidic compartment by E64 and Z-Phemarkers are shown on the left. Note that the 293A10 line expressed Ala-CHNz-insensitive (noncysteine) proteases, andfurther large amounts of immature (-110 kDa) and secreted (-120 kDa) degraded by endosomal-lysosomal cysteine proteases sensitive APP, and that a second larger species was found in the medium. In to these two inhibitors. contrast, the efficiency of secretion of truncated APP was considerTo further evaluate the effects of lysosomal protease inhiably less in the 293A14 line and was accompanied the by accumulation bition on APP processing, we quantified the activities of of detectable levels of fully mature APP (-130 kDa) in the cells. several lysosomal proteases following treatment with Z-PheAla-CHN2under conditions that blocked degradation of APP Additionally, the 293A10 line secreted lesser amounts of a COOH-terminal fragments. Activities of three major lysososecond APP form, -5 kDa larger than the major secreted mal proteinases, the aspartic protease cathepsin D (Takahashi species. In comparison with mock-transfected 293 cells, the level of APP overexpression in the two clones was approxi- and Tang, 1981) and the cysteine proteases cathepsins B and L (Barrett andKirschke, 1981),were quantified 2 h after the mately 10-fold. APP751 Processing in Acidic Compartments of 293A14 addition of inhibitor to 293A14 cells (Fig.3). Z-Phe-Ala-CHN2 Cells-Because the efficiency of APP secretion was lowin the potently inhibited the cysteine proteases cathepsins B and L 293A14 clone, this line was selected for examination of nonse- but slightly increased cathepsin D activity. The concentration cretory pathways of APP processing. First, APP processing of Z-Phe-Ala-CHN2that caused the maximal increase in APP was assessed by labeling with [35S]methioninecoupled with fragments, 1 PM, also produced substantial inhibition of the immunoprecipitation. Two antibodies to domains near the two major lysosomalcysteine proteinases. A Noncysteine Protease Generates LargeNumbers of PotenAPP COOH terminus were used AblOE, directed at the 24 tially AmyloidogenicFragments, while CysteineProteases Furresidues immediately downstream of the transmembrane domain (APP705-728); and AblZE, directed at the COOH- ther Degrade Them-The results presented above suggestthat terminal 30 residues (APP722-751). Cells were treated with APP is processed within an acidic compartment by noncythe acidotropic agents ammonium chloride or chloroquine, steine proteases to generate several 10-22-kDa COOH-terand APP COOH-terminal fragments were isolated by immu- minal fragments, which are then furtherdegraded by lysosonoprecipitation, separated by high resolution Tris/Tricine mal cysteine proteases. If this hypothesis is correct, then SDS-PAGE (Schagger and vonJagow, 1987; Estus et al., cysteine protease inhibition should not influence the rate of 1992), and detected by fluorography. Untreated cells con- appearance of the COOH-terminal fragments but should slow tained a single prominent 9-kDa APP COOH-terminal frag- their rate of disappearance. This was examined by pulsement, and small amounts of fragments of 10,12, 13,19, and labeling 293A14 cells for30 min, followed bychasing for times 22 kDa(Fig. 2, control). AblO, raised against a different ranging between 0 and 180 min (Fig. 4). In untreated cells, COOH-terminal domain, immunoprecipitated the same frag- APP COOH-terminal fragments of 8.5, 9, 10, 12, 13, 15.5, 19, ments (data not shown), while the fragments were not ob- 22,23, and 30 kDa became detectable at 15 min of chase, served when the immunoprecipitation was performed without reached maximal levels at 30 min, and declined thereafter, AblO or 12 (NO 1”Ab) and were greatly reduced in untrans- with apparent tl,+ranging from 30 to 60 min (Fig. 4B). Intact fected 293 cells, indicating that the fragments are derived APP was present at high levels from0 to 60 min of chase and from APP. Acidotropic agents did not decrease the level of then declined with a tl,+of about 30 min.Following a 2-h the 9-kDa APP fragment but instead had little effect (am- treatment with E64, the synthesis and degradation of intact monium chloride) or increased it (chloroquine). In contrast APP were unaffected. Similarly, the appearance of COOH-

-

-7-


FIG.2. APP processing in acidic compartments by both cysteine and noncysteine proteases. APP COOHterminal fragments were identified in 293A14 cells by metabolic labeling and immunoprecipitation as described under "Materials and Methods." Acidotropic agents and protease inhibitors were added to cells 30 min prior to labeling. Shown here is a fluorogram of a 15% Tris/TricineSDS gel. Molecular mass markers are shown on the right. Polypeptides of 9, 10, 11, 13, 15, 17, 19, 22, and 30 kDa were specifically immunoprecipitated by Abl2E. The lysosomotropic agentsNH,Cl and chloroquine reduced the content of the 10-30-kDa APP COOH-terminal fragments (lysosomal) but had little effect on the 9-kDa fragment (secretory). In sharp contrast, the cysteine protease inhibitors E64 and ZPhe-Ala-CHN2 increased the content of the 9-30-kDa fragments. However, the protease inhibitors had no effect when combined with the acidotropic agents.

16605

46

30

, .

L

'

21

14

65

3.4

I I

5

v

L

0

200

600 400

[Z-Pho-AIa-CHN2]

800

1000

FIG.3. Inhibition of lysosomal cysteine proteases of 293A14 cells by Z-Phe-Ala-CHN,. Cells were treated for 2 h with the indicated concentrations of Z-Phe-Ala-CHN2, then were washed to remove inhibitor, and the residual activities of the lysosomal proteases cathepsins B, D, and L were determined as described under "Materials and Methods."

1200

(nM)

terminal derivatives, measuredat a short chase time, was not influenced by E64 (Fig. 4,A and B ) , indicating that cysteine proteases are not responsible for the initial degradation of APP and the genesis of these fragments. Instead, E64 treatment caused a marked slowing in the rate of disappearance of all the 8.5-30-kDa fragments (Fig. 4). This was particularly evident at 60-120 min of chase, at which time the fragments persisted in the presence of E64 but had rapidly disappeared in its absence.None of the polypeptideswasobserved in immunoprecipitationslacking APP antibody (NOI'Ab). It is estimated that E64 prolonged the half-lives of the COOHterminal derivatives by 2-4-fold. Lysosomal Cysteine Protease Inhibition Leads to Increased

Steady-state Levels of Potentially Amyloidogenic APP Fragments-The hypothesis that noncysteine proteases in an acidicorganelle generate APP COOH-terminal fragments, whilelysosomalcysteine proteases further degradethem, leads to the prediction that chronic cysteine protease inhibition shouldcause a steady-state accumulation ofCOOHterminal fragments. To test this prediction, 293 cells or the APP751 overexpressing line 293A14 were treated for 1.5 days with E64 (100 FM) or Z-Phe-Ala-CHNz(1p ~ ) Steady-state . levels of APP and itsCOOH-terminalderivatives were determined by immunoprecipitation with AblZE,followed by separation by Tris/Tricine SDS-PAGE and immunoblot detection with Abl2E or -10E. Neither E64 nor Z-Phe-Ala-CHNz

16606

APP Degradation in Acidic Compartments A

CHASE(YIN):

0

60

30

15

120

*Ai-++& NO P A b

- + - + - + - + - + +

E64

I Mrx100

FFIAGYENTS

FIG. 4. Effects of cysteine protease inhibitor on the rates of appearance and disappearance APP of COOH-terminal fragments. 293A14 cells were labeled with [35S]methionine for 30 min, then washed and chased with medium containing unlabeled methionine. APP and itsCOOH-terminal fragments were isolated by immunoprecipitation with AblZE, except where indicated (NO I'Ab), separated on 15%Tris/ Tricine SDS gels, identified by fluorography, and quantified by densitometry. When present, E64 (100 PM) was added 2 h prior to the onset of labeling. A , fluorogram illustrating the appearance and disappearance of COOH-terminal fragments (top, long exposure) and intact APP(bottom, short exposure). Note that at15-30 min of chase, E64 had little effect on the content of intact APP or its COOH-terminal fragments. However, a t 60-120 min, E64 greatly increased the content of 9-30-kDa APP COOH-terminal fragments. B , quantitative analysis of the effects of E64 on intact APP and the 9-, 13-, 22-, and 30-kDa fragments. Open boxes, no inhibitor; filled boxes, +E64. The contentof APP and its COOH-terminal fragments is inarbitrary units. Note that cysteine protease inhibition did not decrease the rate of production of APP fragments or alter the content of intact APP but instead slowed the rate of disappearance of the fragments.

INTACT APP

I

9 kDa

"

0 0

30

60

90

120

150

180

CHASE (MIN)

CHASE (MIN)

13 kDa

0

30

60

90

---

I

22 kDa

120 180150

CHASE (MIN)

CHASE (MIN)

1

30 kDa

CHASE (MINI

affected steady-state levels of intact APP (Fig. 5, top of gel), consistent with the initial degradation of APP being mediated by a noncysteine protease. However, E64 and Z-Phe-AlaCHNz treatments caused massive increases in at least 13 immunoreactive polypeptides ranging from 8.5 to 22 kDa. The

13 polypeptides likely are COOH-terminal fragments of APP because: 1) they are immunoprecipitated by 12E and stained by either 12E (Fig. 5 A ) or 10E (Fig. 5 B ) , antibodies to two distinct APP domains near the COOH terminus; 2) they are not immunoprecipitated in the absence of primary antibody


t

16607

s 46 30

i

21

COOH-TERMINAL FRAGMENTS

14

6.5

34

293

293A14

293A14293

293A14

FIG.5. Chronic inhibition of lysosomal cysteine proteases causes accumulation of at least 12 APP COOH-terminal fragments. Either wild-type 293 or 293A14 cells were treated for 1.5 days with E64 (100 p ~or)Z-Phe-Ala-CHN2(1 p ~ ) APP . COOH-terminal fragments were isolated by immunoprecipitation (with AblZE), separated on 15% Tris/Tricine SDS gels, and detected by immunoblotting with AbslOE or - 12E as described under "Materials and Methods." A and C, blots stained with Ab12E; B, blot stained with AblOE. Molecular weight markers are shown in A on the right. Note that both 1OE and 12E label a t least 12 polypeptides ranging from 8.5 to 22 kDa in 293A14 cells treated with protease inhibitor, but not in the absence of protease inhibitor or in untransfected 293 cells. 50- and 25-kDa polypeptides were labeled nonspecifically, since they were present in untransfected cell lysates, in cells not treated with protease inhibitor, and when the immunoprecipitation was performed without the 12E Ab ( C , NO PAb) or with Ab12E preabsorbed with excess peptide immunogen (C, preabsorbed). For size comparison, recombinant APP652-751 generated in a bacterial expression system is shown (C). Note that atleast 9 of the APP fragments (13-22 kDa) are aslarge or larger than APP652-751. or with 12E preabsorbed with an excess of the peptide immunogen (Fig. 5C); 3) their content is greatly reduced in 293 cells not transfected to overexpress APP751 (Fig. 5, A and B). In untransfected 293 cells, cysteine protease inhibition also led to detectable increases in the 9- and 13-kDa derivatives, but the content of these fragments was substantially less than in the 293A14 transfectant. In overexpressing 293A14 cells, the contentof APP COOHterminal derivatives increased at least 10-fold upon cysteine protease inhibition; this estimate is a minimum value since, in untreated cells, most fragments were below the limit of detection. In addition to COOH-terminal fragments of 8.5,9, 10, 11, 13, 15.5, 19, and 22 kDa detected previously by metabolic labeling procedures,derivatives of 16, 18.5, and 20 kDa also accumulated with cysteine protease inhibition. The 13kDa fragment co-migrated with a recombinant APP segment corresponding to APP652-751,which starts 1 residue upstream of the NH2 terminus of the PIA4 domain (Fig. 5C). Therefore, the 13-30-kDa fragments identified in Figs. 2, 4, and 5 may contain the entire PIA4 domain and therefore could bepotentially amyloidogenic. As another test of the hypothesis that lysosomal cysteine proteases are involved in the clearance, but not the genesis, of potentially amyloidogenic APP fragments, immunohistochemistry with Abl2E was used to localize APP COOHterminal epitopes after E64 treatment. In untreated 293A14 cells, APP immunoreactivity was diffusely distributed throughout the cytoplasm and was concentrated in a perinuclear cluster, probably corresponding to endoplasmic reticulum-Golgi (Fig. 6A). Following treatment for 1.5 days with E64, immunoreactivity for COOH-terminal epitopes of APP (detected with Ab12E) redistributed within the cytoplasm. Intense punctate immunoreactivity in the form of small spheres filled all cells, although staining was excluded from

FIG.6. Cysteine protease inhibitor causes accumulationof APP COOH-terminal epitopes within lysosomes.293A14 cells were stained for APP-like immunoreactivity ( A and B ) or cathepsin D-like immunoreactivity ( C ) . A , untreated cells; B and C, treated with 100 p~ E64 for 1.5 days prior to immunostaining. Note that in untreated cells, APP COOH-terminal epitopes were concentrated in a perinuclear cluster. Upon E64 treatment, APP COOH-terminal epitopes were concentrated in small spheres that filled the cytoplasm ( B )and had a localization similar to cathepsin D ( C ) .Scak bar = 25 prn.

the nucleus (Fig. 6B). Comparably treated cells labeled with an antibody to cathepsin D, a marker of secretory vesicles, late endosomes, and lysosomes (Rijnboutt et aL, 1992), exhibited staining of intracytoplasmic puncta in a manner similar to the APP antibody (Fig. 6C). Thus, chronic cysteine protease inhibition causes an accumulation of APP COOH-terminal epitopes within presumptive endosomes-lysosomes.


16608 DISCUSSION

In human 293 cells,we have provided evidence that APP751 is degraded within an acidic compartment to produce at least 12 fragments, 9 of which are potentially amyloidogenic, and that these fragments are, in turn, rapidly degraded in Iysosomes. Moreover, there is a segregation of protease function; a noncysteine protease (or proteases) initiates APP degradation in an acidic compartment to produce potentially amyloidogenic fragments, while the lysosomal cysteine proteases are responsible for their clearance. The findings are compatible with previous observations linking lysosomes with APP degradation (Benowitz et al., 1989; Cole et al., 1989; Golde et al., 1992; Caporaso et al., 1992; Haass et al., 1992a) and possibly with @/A4formation (Cataldo et al., 1990, 1991; Burdick et al., 1992) and begin to identify specific proteolytic events that may be important for amyloidogenesis. Considerable clonal variation in 293 cells was observed in the trafficking of APP751. Among the 60 stably transfected clones examined, the efficiencies of APP secretion and mechanisms of APP processing differed markedly. This variation has been exploited in two ways.First, cells such asthe 293A14 line, in which secretory processing of APP is relatively inefficient (Fig. l),are ideally suited for analyses of nonsecretory pathways of APP trafficking. Second, cells such as the293A10 line produce APP forms that have not previously been described, including a second secreted species -5 kDa larger than the major form. Should this be a secreted APP form extended at the COOH terminus, it would likely contain the entire @/A4domain and represent a potentially amyloidogenic soluble APP. The polypeptide composition and route of genesis of the minor secreted form are currently under investigation. An indication of the segregation of protease function inthe processing of APP751 comes from the accumulation of APP fragments in response to E64, a general cysteine protease inhibitor(Barrett et al., 1982), or to 2-Phe-Ala-CHN,,a selective andpotent irreversible blocker of the lysosomal cysteine proteases cathepsins B and L that does not inhibit the nonlysosomal cysteine proteases, the calpains, or proteases of other classes (Green and Shaw, 1981; Kirschke and Shaw, 1981; Crawford et al., 1988). Previous studies have examined the effect of the protease inhibitor leupeptin on APP processing (Cole et al., 1989; Goldeet al., 1992; Haass et al., 1992a) and have suggested that leupeptin-sensitive proteases may be responsible both for the generation and the elimination of APP COOH-terminal fragments (Golde et ol., 1992). The present use of irreversible and specific cysteine proteaseinhibitorspermitteda parallel evaluation of the effects of inhibitortreatment on the activities of specific lysosomal proteases (Fig. 3) and on APP processing (Figs. 2, 4, and 5). If a cysteine protease were important for initiating endosomal-lysosomal APP degradation, then its inhibition should block the appearance of APP fragments and, perhaps, cause accumulation of intact APP. Instead, thecysteine protease blockers did not modify intact APP levels and caused a marked increase, rather than a decrease, in the content of derivatives. This was apparent with short-term cysteine protease inhibition (3 h) using a metabolic labeling and immunoprecipitation procedure (Fig. 2),as well as with longer term inhibition (1.5 days), which led to massive increases in the steady-state levels of at least 12 fragments (Fig. 5 ) . Pulsechase experiments confirmed that cysteine protease inhibition did not alter the appearance of APP fragments but instead slowed their rates of disappearance (Fig. 4). In theabsence of protease inhibitors, all of the COOH-terminal APP fragments were rapidly cleared, with half-lives ranging from about 30 to

60 min. These results indicatethat thecysteine protease class is responsible for rapid degradation of APP COOH-terminal derivatives. The inhibitorconcentrations that caused maximal increases in APP fragments also produced substantial blockade of lysosomal cathepsinsBand L but spared the aspartic protease cathepsin D. Activities of cathepsins B and L were determined using highly preferred substrates (Barrett and Kirschke, 19811, but it cannot be completely ruled out that some of the activity ascribed to cathepsin L may be partly due to another recently cloned member of the cysteine protease family, cathepsin S (Wiederanders et al., 1992). Additional inhibitors with greater selectivity would be useful for defining the roles of cathepsins B, L, and S in the clearance of potentially amyloidogenic APP fragments. There is considerable evidence that endosomes-lysosomes are sitesfor the elimination of potentially amyloidogenic APP derivatives, but the identity of the acidic compartment in which APP degradation is initiated and potentially amyloidogenic fragments are formed is less clear. The acidotropic agents ammonium chloride and chloroquine reduced the content of 10-22-kDa COOH-terminal fragments and prevented the cysteine protease inhibitor-evoked increase in APP fragments rangingfrom 10 to 30 kDa (Fig. 2), consistentwith the initial APP degradation and formation of the 10-30-kDa fragments occurring within an acidic compartment. Our immunohistochemical analysis and protease inhibitor studies provide further evidence that the elimination of APP fragments by cysteine proteases takes place in endosomes-lysosomes (Figs 5 and 6). However, these data do not conclusively identify the acidic compartment in which APP degradation is initiated andpotentially amyloidogenic fragments areformed. While APP degradation could be initiated in endosomeslysosomes, it isalso possible that theacidotropic agents could inhibit the formation of COOH-terminal derivatives by interfering with other compartments known to maintain an acidic environment, such as the trans-Golgi or its associated secretory vesicles (Anderson and Pathak, 1985). Previous studies have used acidotropic agents (Cole et al., 1989; Golde et al., 1992; Caporaso et al., 1992), isolation of lysosome-enriched subcellular fractions, and surface labeling coupled with immunahistochemistry (Haass et al., 1992a) to identify endosomes-lysosomes as a site of degradation of APP derivatives but also have not distinguished between the many intracellular acidic compartments as sites for the formation of APP derivatives. The possible involvement of lysosomes in amyloidogenesis is supported by the finding that at least 9 APP COOHterminal fragments are routinely degraded within lysosomes that, on the basis of their size in comparison with a recombinant APP segment, likely contain the entire PIA4 domain (Fig. 5). The processing of potentially amyloidogenic APP fragments within acidic compartments such as lysosomes is probably not a peculiarity of the 293A14 cell line, because some similar, if not identical, COOH-terminal fragments have been isolated from human brain homogenates (Estus et al., 1992), microvessel preparations (Tamaoka et a i , 1992), and endothelial cells (Haass et al., 1992a). Recently, production and secretion of @/A4 protein has been described in several types of cells in culture (Haasset al., 1992b;Shoji et al., 1992), but the degradative pathway leading to this @/A4formation has not been elucidated. If an acidic compartment is involved in @/A4formation, the degradation of potentially amyloidogenic fragments in lysosomes predominantly by the cysteine proteases suggests that thisprotease class may beresponsible for amyloidogenesis. An alternative possibility may be that

A P P Degradation i n Acidic Compartments @/A4 is generated directly from APP by a noncysteine protease, one that may not require the intermediate formationof the potentially amyloidogenic fragments detected in the present study. In a third possibility involving an acidic compartment, amyloidogenic APP fragments derived by the action of an acid-preferring noncysteine protease may be further processed in another compartmentto form @/A4.In thisinstance, cytoplasmic or extracellular proteases could be important for @/A4 production. In the latter two scenarios, lysosomal cysteine proteases would serve the beneficial role of eliminating potentially amyloidogenic fragments without forming @/A4 protein. The availability of model systems that exhibit @/A4 formation should allow direct evaluation of the involvement of specific cellular compartments in amyloidogenesis. Acknowledgments-We thank Drs. J. Kauer, M. Iqbal, and the peptide chemistry group a t Cephalon for providing peptides for antibody production, and Schering-Plough Corporation and Drs. F. Baldino and J. Vaught for their continued support of this research. REFERENCES Anderson, R. G. W., and Pathak, R. K. (1985)Cell 40,635-643 Anderson, J. P., Esch, F. S., Keim, P. S., Sambarmurti, K., Lieberburg, I., and Robakis, N. K. (1991)Neurosci. Lett. 128, 126-128 Barrett, A. J., and Kirschke, H. (1981)Methods Enzymol. 80, 535-561 Barrett, A. J., Kembhavi, A.A., Brown, M. A,, Kirschke, H., Knight, C. G., Tamai, M., and Hanada, K. (1982)Biochem. J. 201,189-195 Benowitz, L. I., Rodriguez, W., Paskevich, P. Mufson, E. J., Schenk, D., and Neve,R.L. (1989)Ex Neurol. 106,237-i50 Burdick, D., Soreghan, If;, Kwon, M., Kosmoski, J., Knauer, M., Henschen, A,, Yates, J., Cotman, C., and Glabe, C. (1992)J. Biol. Chem. 267,546-554 Ca oraso, G. L., Gandy. S. E., Buxbaum, J. D., and Greengard, P. (1992)Proc. fiatl. Acad. Sci. U. S. A . 89,2252-2256 Castano, E.M., Ghiso, J., Prelli,F., Gorevic, P. D., Migheli, A,, and Frangione, B. (1986)Biochem. Bio hys Res Commun. 141,782-789 Cataldo, A. M., Thayer, Y.1 Bird, E. D., Wheelock, T. R., and Nixon, R. A. (1990)Brain Res. 613, 181-192 Cataldo, A. M., Paskevich, P. A,, Kominami, E., and Nixon, R. A. (1991)Proc. Natl. Acad. Sei. U. S. A. 88, 10998-11002 Cole, G. M., Huynh, T. V., and Saitoh,T. (1989)Neurochem. Res. 14,933-939 Crawford, C., Mason, R. W., Wikstrom, P., and Shaw, E. (1988)Biochem. J. 263,751-758 Dyrks, T., Weidemann, A,, Multhaup, G., Salbaum, J. M., Lemaire, H.-G., Kang, J., Muller-Hill, B., Masters, C. L., and Beyreuther, K. (1988)EMBO J . 7.949-957 - , - ---- Esch, F. S., Keim, P. S., Beattie, E. C., Blacher, R. W., Culwell, A. R., Oltersdorf, T., McClure, D., and Ward, P. J. (1990)Science 248, 1122-1124 Estus, S., Golde, T. E., Kunishita, T., Blades, D., Lowery, D., Eisen, M., Usiak,

8.

~

16609

M., Qu, X., Tabira, T., Greenberg, B. D., and Younkin, S. G. (1992)Science 266,726-728 Glenner, G. G., and Wong, C. W. (1984)Biochem. Biophys. Res. Commun. 120, ms-xqn -_" Golde, T. E., Estus, S., Younkin, L. H., Selkoe, D. J., and Younkin S. G. (1992) Science 266,728-730 Green, G. D. J., and Shaw, E. (1981)J. Biol. Chem. 266, 1923-1928 Haass. C.. Koo. E. H.. Mellon. A.. Hune. A. Y.. and Selkoe. D. J. (1992a)Nature 367,5&-503 Haass, C., Schlossmacher, M. G., Hung, A.Y., Vigo-Pelfrey, C., Mellon, A., Ostaszewski, B.L., Lieberburg, I., Koo, E. H., Schenk, D., Teplow, D. B., and Selkoe, D. J. (1992b)Nature 369,322-325 Kang, J., Lemaire, H. G., Unterbeck, A., Salbaum, J. M., Masters, C. L., Gneschick, K. H., Multhaup, G., Beyreuther, K., and Muller-Hill, B. (1987) Nature 326,733-736 Kirschke, H., and Shaw, E. (1981)Biochem. Biophys. Res. Commun. 101,454458 Kirschner, D. A., Inouye, H., Duffy, L. K., Sinclair, A., Lind, M., and Selkoe, D. J. (1987)Proc. Natl. Acad. Sci. U. S. A . 84,6953-6957 Kitaguchi, N., Takahashi, Y., Tokushima, Y., Shiojiri, S., and Ito, H. (1988) Nature 331,530-532 Maruyama, K., Terakado, K., Usami, M., and Yoshikawa, K. (1990)Nature 347,566-569 Masters, C. L., Simms, G., Weinman, N. A,, Multhau , G , McDonald, B. L.. and Beyreuther, K. (1985)Pmc. Natl. Acad. Sei. U. ?L A. '82,4245-4249 Neve, R. L., Kammescheidt, A,, and Hohmann, C. F. (1992)Pmc. Natl. Acad. scl. u. A. 89,3448-3452 Nordstedt, C., Gandy, S. E., Alafuzoff, I., Caporaso, G. L., Iverfeldt, K.,Grebb, J. A., Winblad, B., and Greengard, P. (1991)Proc. Natl. Acad. Sa. U. S. A. 88,8910-8914 Oltersdorf, T., Ward, P. J., Henriksson, T., Beattie, E. C., Neve, R., Lieberburg, I., and Fritz, L. C. (1990)J. Bwl. Chem. 266,4492-4497 Ponte, P., Gonzalez-DeWhitt, P., Schillin J , Miller, J., Hsu, D., Greenberg, B., Davis, K., Wallace, W., Lieberburg, Fuller, F., and Cordell, B. (1988) Nature 331,525-527 Rijnboutt, S., Stoorvogel, W., Geuze, H. J., and Strous, G. J. (1992)J. Biol. Chem. 267, 15665-15672 Savage, M. J., Iqbal, M., Loh, T., Trusko,S. P., Scott, R., and Siman,R. (1993) Neuroscience. in Dress Scha ger, H , and ;on Jagow G. (1987)A d . Biochem. 166,368-379 Schugert, D:, LaCorbiere, M.: Saitoh, T., andCole, G. (1989)Proc. Natl. Acad. Sci. U. S. A. 86,2066-2069 Selkoe, D. J. (1991)Neuron 6,487-498 Shoji, M., Golde, T. E., Ghiso, J., Cheung, T. T., Estus, S., Shaffer, L. M., Cai, X.-D., McKay, D. M., Tintner, R., Frangione, B., and Younkin, S. G. (1992) Science 268,126-129 Siman, R., Card, J. P., and Davis, L. G. (1990)J. Neurosci. 10,2400-2411 Takahashi, T., and Tang, J. (1981)Methods Enz mol 80,565 581 Tamaoka, A,, Kalana, R. N., Lieberburg, I., and &elk& D. J. (1992)Proc. Natl. Acad. Sci. U. S. A. 89, 1345-1349 Tanzi, R. E., McClatchey, A. I., Lampert, E. D., Villa-Komaroff, L., Gusella, J. F., and Neve, R. L. (1988)Nature 331,528-530 Weidemann, A., Konig, G., Bunke, D., Fischer, P., Salbaum, J. M., Masters, C. L., and Beyreuther, K. (1989)Cell 67,115-126 Wiederanders, B., Bromme, D., Kirschke, H., v. Figura, K., Schmidt, B., and Peters, C. (1992)J. Biol. Chem. 267, 13708-13713 Yankner, B. A., Dawes, L. R., Fisher, S., Villa-Komaroff, L., Oster-Granite, M. L., and Neve, R. L. (1989)Science 246,417-420 "

,

I

,

'

s.

I.,