Secretion of &Amyloid Precursor Protein Involves Multiple Cleavage ...

17 downloads 76732 Views 2MB Size Report
ized the proteolytic cleavage site for D-APP secretion by amino acid sequence analysis of the -9-kDa B-APP carboxyl-terminal cleavage product produced by re-.
Vol. 269, No. 1, Issue of January 7, pp. 627-632, 1994 Printed in U.S.A.

THEJOURNAL OF BIOLOGICAL CHEMISTRY

0 1994 by The American Society for Biochemistry and Molecular Biology, Inc.

Secretion of &Amyloid Precursor ProteinInvolves Multiple Cleavage Sites* (Received for publication, July 7, 1993, and in revised form, August 9, 1993)

Ziyang Zhong, Jeffrey Higaki, Kenji MurakamiS, Yu Wang, Rosanne Catalano, Diana Quon, and Barbara Cordells From Scios Nova Inc., Mountain View,California 94043

A major histopathological feature of Alzheimer's dis- brane spanning domain followed by a short cytoplasmic tail ease is deposits of a -4-kDa &amyloid peptide derived (1,5-7). All /3-APPisoforms are proteolytically cleaved within by proteolytic processing from a precursor, theB-amy- the @-amyloidsequence to generate a large soluble aminoloid precursor protein(B-APP). Proteolytic cleavage of terminal derivative of p-APP which is secreted and a mem&APP within the -4-kDa &amyloid domain permits brane-bound carboxyl-terminal fragment of-9 kDa (8-10). the secretion of the amino-terminal portion of @-APP The secretion of soluble ,8-APP, extensively documented in while concomitantly producing a membrane bound -9- vitro, appears also to occur in vivo based on the presence of kDa carboxyl-terminal fragment. We have character- carboxyl-terminal truncated ,6-APP in cerebrospinal fluid (6, ized the proteolytic cleavage site for D-APP secretion 11) and serum? It has been determined that a constitutive by amino acid sequence analysis of the -9-kDa B-APP secretory pathway produces soluble P-APP (12) andthat carboxyl-terminal cleavage product produced by recombinant and native expression systems. Recombi- proteolytic cleavage appears to occur only with fully glycosylocalizing cleavage to the nant 8-APP was generatedby a vaccinia virus expres- lated precursor protein (13), thereby trans-Golgi, to a secretory vesicle, or to the cell membrane sion system in CV-1 monkey fibroblasts; endogenous surface. Evidence for both intracellular cleavage and mem&APP was obtained using a fibroblast line derived from an individual with Down's syndrome. The se- brane surface cleavage has been reported (13-15). Esch et al. quences of both unlabeled and metabolically radiola- (8) and Sisodia et al. (9) were the first to demonstrate that beled -9-kDa fragment fromCV-1 cells reveal that the the cleavage releasing soluble P-APP occurs within the pamyloid peptide domain. In the report by Esch et al., the major (60%)secretory cleavage site is after Lys" of the @-amyloiddomain as reported previously; however, cleavagewas found to occur after Lys"of the p-amyloid an additional cleavage site is seen after Phe" (40%). sequence. Since these original observations, the secretory Radiosequence analysis of the carboxyl-terminalfrag- cleavage site of @-APPhas been studied in a variety of ment purified from Down's syndrome fibroblasts in- mammalian cultured cell systems (15-19), in cultured insect dicates cleavage sites after Phe", G1uZ2, and GlyZK and cells (20, 21), and in Alzheimer's disease brain (22). not after Lys". CV-1 cells expressing B-APP mutants Several different approaches have been employed to define lacking 4 and 6 amino acids adjacent to Lys" yielded the /3-APP secretory cleavage event. These approaches in-9-kDa fragments with two identical cleavage sites, cluded characterization of the carboxyl terminus of secreted neither of which occurred after the retainedLys" but 8-APP (8, 18-20, 22), amino-terminal analysis of the -9-kDa were after Glu" and Hid3. These data suggest that fragment generated upon cleavage (8,17,21), andsite-specific secretion of &APP involves multiple proteinases and mutagenesis of the secretory cleavage site (15-17). In the that the composition of these proteinases may vary direct biochemical studies with isolated protein, only the within differentcell backgrounds. previously identified Lys" cleavage site wasobserved. In experiments in which p-APP was mutated at and around this cleavage site, @-APPsecretion was seen with all mutants. The deposition of @-amyloidin the brain parenchyma and Although the cleavage site in the mutantswas not specifically cerebral vasculature is a histopathological characteristic of defined (except for one mutant), it has been concluded that Alzheimer's disease. @-Amyloidis a 39-42-amino acid peptide the cleaving proteinase, referred to as P-APP secretase, lacks that results from proteolytic processing of a larger precursor, residue or sequence specificity but instead cleaves at a fixed the P-amyloid precursor protein (p-APP).' Three major iso- distance from the membrane (15-17). The primary goal of forms of B-APP encoding proteins of 695,751, and 770 amino this study was to address theapparent lack of sequence acids have been described (1-4). p-APP isoforms are highly specificity by P-APP secretase. In thisreport, we examine the modified membrane glycoproteins with a single transmem- cleavage event in greater biochemical detail and show that multiple cleavage sites are used for both native and recombi* This work was supported by Marion Merrell Dow, Inc. The costs nant P-APP751 secretion, indicating that a diverse set of of publication of this article were defrayed in partby the payment of proteinases may be responsible for the secretion of this molpage charges. This article must therefore be hereby marked "aduertisement" in accordance with 18U.S.C. Section 1734 solelyto indicate ecule. this fact. Present address: Daiichi Pharmaceutical Co., 16-13 Kita-Kasai 1-Chome, Edogawa-ku, Tokyo Japan. To whom all correspondence should be addressed Scios Nova Inc., 2450 Bayshore Parkway, Mountain View, CA 94043. The abbreviations used are: p-APP, P-amyloid precursor protein; DS, Down's syndrome; VV, vaccinia virus; Tricine, N-[P-hyroxy-l,lbis(hydroxymethyl)ethyl]glycine.

EXPERIMENTALPROCEDURES

Cell Culture, Drug Treatment, Radiolabeling, and Virus InfectionCV-1 cells, an African monkey kidney fibroblast line, were obtained from the American Type Culture Collection (Rockville, MD) as were

627

D. Quon and B. Cordell, unpublished observations.

628

P-APP Secretase Cleavage Sites

the COS-7 cells. DS fibroblasts, a line derived by skin biopsy from with tritium,a sample labeled with [36S]methionine/cysteinewas an individual documented to be trisomic for chromosome 21, were electrophoresed in an adjacent lane on the gel. The region on the obtained from the Human Genetic Mutant Cell Repository Coriell polyvinylene difluoride membrane containing the -9-kDa fragment Institute for Medical Research (Camden, NJ) under the repository was excised for amino-terminal aminoacid sequence analysis. number GM02504G. Cells were propagated as suggested by the proLarge Scale Preparation of -9-kDa fragment-For large scale providing agencies. Briefly, cell culture conditions were as follows: CV-I duction of -9-kDa fragment, -IO9 CV-1 cells grown in roller bottles in Eagle's minimum essential medium with 10% fetal bovine serum, (-5 X lo7 cells/bottle) were infected with VV:751, lysed, and the COS-7 cells in Dulbecco's modSed Eagle's medium-21 with 10% fetal fragment partially purified by ion exchange chromatography. Cell bovine serum, and DS fibroblasts in Eagle's minimun essential me- lysates were diluted with an equal volume of 50 mM Tris-HC1, pH dium with 20% fetal bovine serum. Infections with 8-APP recombi- 7.5,150 mM NaCl, 1%sodium deoxycholate, 1%SDS) and 10 volumes nant vaccinia viruses were carried out as described previously (23) in of 10 mM Tris-HC1, pH 7.5. The mixture was loaded onto a 50-ml which each titred virus stock was used to infect washed CV-1 cell bedvolume of DEAE Sepharose CL-GB (Pharmacia) equilibrated monolayers at a multiplicity of infection of 0.1 in Eagle's minimum with 10 mM Tris-HC1, pH 7.5. After loading the sample, the column essential medium containing 0.2% fetal bovine serum. Following a 2- was washed with 10 mM Tris-HCI, pH 7.5, and thenwith 10 mM Trish adsorption period, monolayers were washed, and a 10-fold excess HCl, pH 7.5, 100 mM NaCl. The -9-kDa fragment was eluted with of serum-free mediumwas added for the remainder of the 16-h 10 mM Tris-HC1, pH 7.5,200 mM NaCl. The column eluent was infection period. For experiments involving radiolabeling of 8-APP, immunoprecipitated with 369Ab and protein A-Sepharose. Fragment cultures were washed then incubated for 30 min in serum-free medium isolation was carried out by SDS-polyacrylamide gel electrophoresis lacking the amino acid(s) to be used in labeling, incubated with either and polyvinylene difluoride membrane transfer as described above 200 &i/ml [%]methionine/cysteine (ICN), 1mCi/ml [3H]lysine,or coupled with staining by Coomassie Blue dye. [3H]isoleucine(Amersham Corp.) for 2 h. In experiments using chloAmino-terminal SequenceAnalysk"olyviny1ene difluoride memroquine treatment, 100 pg/ml of drug was applied 30 min prior to the branes containing -9-kDa protein were subjected to Edman degraaddition of radiolabel and was present during the entire labeling dation using an Applied Biosystems A77A protein sequenator (Foster period. City, CA) using the BLOTT-2 program according to the manufacturPreparation of 8-APP Mutant Viruses-Mutants of P-APP695 er's recommendations. For unlabeled protein, amino acid analysis were constructed using oligonucleotide-directed site-specific muta- was achieved by an on-line Applied Biosystems model 120A phenylgenesis according to standard procedures. Briefly, an EcoRI-PuuII thiohydantoin derivative analyzer coupled to the sequenator. For fragment encoding the carboxyl-terminal 103 amino acids of 8-APP radiolabeled protein, each degradation cycle wascollected, mixed with was cloned into pBluescript I1 (KS-) vector (Strategene, La Jolla, scintillation fluid, and then counted in a Beckman LS 3801 scintilCA). Single-stranded DNA was isolated after transfection of DNA lation counter. into JMlOl bacteria with helper phage then used for mutagenesis. Oligonucleotides used to create Lys" replacement mutants were as RESULTS follows. To define the cleavage event allowing secretion of @-APP, LYS" to Leu: 5"AAAGAACACCAACAGTTGATGATGACC-3'. we used an African monkey kidney fibroblastic cell line, CVLys" to Ala: 5'-AAAGAACACCAAGGCTl'GATGATGAAC-3'. Lys" to Gln: 5'-AAAGAACACCAACTGTl'GATGATGAAC-3'. 1, and a fibroblast cell line established by skin biopsy from LYS" to Glu: 5'-AAAGAACACCAACTCTTGATGATGAAC-3'. an individual with Down's syndrome (DS fibroblasts) as celLys" to Arg: 5'-AAAGAACACCAAGCGTTGATGATGAAC-3'. lular systems for study and comparison. The DS fibroblast Oligonucleotidesused to create deletion mutants were as follows. line was selected since all individuals with Down's syndrome ALys'? 5'-CAAAGAACACCAATTGATGATGAACTT-3'. AGln16Lys? 5'-CAAAGAACACCAAATGATGAACTTCATA-3'. developAlzheimer's disease. The CV-1 cells were used to express the recombinant 751-amino acid isoform of P-APP AHi~'~-~s''-Gln~~-L 5'-TGCAAAGAACACCAAAACTTCAys~~: (j3-APP751) using a vaccinia virus vector harboring the huTATCCTGA-3'. AGl~~~-Val'~-His'~-His"-Gln'~-Lys~~: 5"TGCAAAGAACACCAman cDNA (VV751). The vaccinia virus expression system AATATCCTGAGTCATG-3'. produces large amounts of recombinant protein (-100-fold ALeu17: 5'-CTGCAAAGAACACTTTTTGATGATGAA-3'. over endogenous levels) (23), thereby facilitating analysis of A L ~ U ' ~ - V ~5'-TCTTCTGCAAAGAATGATGATGAA~'~: the /3-APP secretase cleavage site(s). TheDown's fibroblasts 3'. ALeu'7-Val'S-Phe's-Phem(KM-3): 5'-CCACATCTTCTGCTTTT predominantly express 8-APP751. P-APP from this cell line represents processed native protein. TGATGATGAAC-3'. ALeu"-Va11S-Phe'a-Phe20-Ala2'-Glu21 (KM-4): B'-TTTGAACCCASequence of 8-APP Secretion Cleavage Site($-To characCATCTTTTTGATGATGAAC-3'. terize the secretion site of recombinant P-APP751 produced Double-stranded phage DNA, shown to be correct by DNA se- in CV-1 cells, the -9-kDa membrane-bound carboxyl-termiquence analysis, was prepared for each mutant from which the EcoRIPuuII plasmid was isolated then reinserted into the8-APP695 cDNA nal product of secretase cleavage was studied. Metabolically carried on a pGEM-3 plasmid. An expression plasmid that harbors radiolabeled -9-kDa fragment was obtained by infecting CVthe human 8-actin promoter was engineered for each mutant andfor 1cells with VV751 inthe presence of either [35S]methionine, wild-type B-APP695. Each mutant DNA wascharacterized for release [3H]lysine,or [3H]isoleucine.The fragment was purified from of 8-APP into medium following transfection of COS-7 cells and the infected cell lysates by immunoprecipitation with an Western blotting using an antiserum raised to the extracellular do- antiserum directed to the cytoplasmic domain of P-APP main of B-APP695 (241, the predominating epitope of which maps near the amino terminus of 8-APP. Construction and isolation of (369Ab) followed by SDS-polyacrylamide gel electrophoresis. recombinant vaccinia viruses harboring KM-3 and KM-4 deletion Fig. 1B illustrates the pattern of [Y3]methionine/cysteinemutants were carried out as described for a vaccinia virus harboring labeled P-APP carboxyl-terminal fragments present in the wild type 8-APP751 (VV751). immunoprecipitate from the CV-1 cell lysate which is well Isolation of -9-kDa Carboxyl-terminal Secretase Fragment-Viralabove that from endogenous 8-APP (Fig. lA). Typically the infected CV-1 cells and uninfected DS fibroblasts werelysed and -9-kDa fragment is the major carboxyl-terminal fragment of immunoprecipitated with an antiserum raised to the cytoplasmic domain of 8-APP (369Ab).Cell lysis was performed exactly according a consistent set of fragments reported by a number of investo theprocedure of Gabuzda et al. (25). The resulting detergent layers tigators (10, 14, 17, 26). The carboxyl-terminal fragments containing the membrane fraction and -9-kDa fragment were used larger than the-9-kDa fragment do not appear to be produced for immunoprecipitation with 369Ab. Protein A Sepharose (Phar- by secretase activity but by degradative processes (10, 14). macia LKB Biotechnology Inc.) was used to recover the 369Ab--9Since the carboxyl-terminal half of (3-APP lacks cysteine kDa protein complex. Immunoprecipitated protein was washed exten- residues, labeling of the -9-kDa fragment will be exclusively sively then subjected to electrophoresis on a 16% SDS-Tris-Tricinepolyacrylamide gel. Protein in the gelwas transferred by electro- at methionine residues. When the -9-kDa fragment was isolated and subjected to blotting onto polyvinylene difluoride membrane, and autoradiography was used to locate the -9-kDa fragment. To localize samples labeled amino-terminal radiosequence analysis, 35Ssignals were found

p-APP Secretase Cleavage Sites

629

had two amino termini, one at Leu" of the @-amyloidpeptide and the second at PheZ0(Fig. 2 4 . :--18 koa Direct amino-terminal sequence analysis was conductedon the /3-APP -9-kDa fragment to confirm the observed heterL - 14 ogeneity obtained by radiosequencing. Preparative amounts I of -9-kDa fragment were generated fromVV:751-infected " ' 6 CV-1 fibroblasts. The fragment was purified identically except FIG.1. Immunoprecipitation of [86S]methionine-labeled for the inclusion of an ion exchange chromatographic step -0-kDa and other B-APP carboxyl-terminal fragmentssepa- needed to remove bulk protein in the lysate prior to immurated by 16%SDS-Tris-Tricine-polyacrylamidegel electro- noprecipitation. For two independent preparations, direct phoresis. Lune A , pSCll control virus-infected CV-1 cell lysate; lane amino-terminal microsequence of the -9-kDa fragment again B , VV:751-infected CV-1 cell lysate; lane C, VV:751-infected CV-1 revealed heterogeneity (Table I). Twosequences wereobcell lysate treated with chloroquine; lane D, VV:KM-3-infected CV-1 tained; one sequence initiates with Led7, the other with Phe2'. cell lysate; lane E , VV:KM-4-infected CV-1 cell lysate; lane F, DS From the relative yields of amino acids at each cycle of fibroblasts. The -9-kDa fragment characterized in each sample was the fastest migrating single protein band on the gel. The protein degradation over four cycles, it can bededuced that the cleavage after Lys" represents fragment migrating immediately above the -9-kDa carboxyl-terminal sequence resulting from fragment in the KM-3 and KM-4 samples was not included in roughly 60%of the total-9-kDa fragment population. These sequence analyses. For lanes A-E infection conditions and exposure direct amino-terminal results are in complete agreement with periods were identical. the cleavage sites determined by radiosequence analysis of the -9-kDa fragment. Together, the dataconfirm the presence of -1 16 Mtt at least two secretase cleavage sites in 6-APP751 produced by recombinant means. To address the possibility that theheterogeneity found with the -9-kDa fragment is a result of secretase cleavage and not a combination of secretase plus an intracellular degradation event, we determined the sequence of -9-kDa fragment produced in the presence of chloroquine, a drug that neutralizes acidic intracellular compartments such as lysosomes. It has beenshown that secretion of @-APP is unaltered in the 9 presence of chloroquine (26). CV-1 cells were infected with R 12 VV:751 during which protein was metabolically labeled with [35S]methionine/cysteinein the presence of 100 pg/ml chloroquine. As shown in Fig. lC, although chloroquine changed the abundance of some p-APP carboxyl-terminal fragments, the -9-kDa secretase product was present as the major species. When this -9-kDa fragment was sequenced, 35Sradioactive signals were identical to that seen in the non-chloroquine-treated sample (Fig. 3). Site-specific Mutugenesis of the Cleavage Domuin-To understand better the apparent lack of sequence specificity of secretase, we constructed a set of mutants at and around Lyd6 of P-APP695 using site-direct mutagenesis (Table 11). Replacement of Lys" in the @-amyloid domainby small hydrophobic residues (Ala or Leu), by acidic residues (Gluor Asp), or by an alternative basic reside (Arg) did not block secretion of @-APP into the medium after transfection and transient cycle expression in COS-7 cells (data not shown). In addition, a A

B

C

1

D

E

1

M

r

.-YEVHHQK"LVF"FAEDVCSNKCA Sequencingcycle W: 1 19 16 I2 16

13

12

9

I 1 CLM." 15

TABLE I NH2-terminal microsequencing of the -9-kDa fragment from CV-I cells infected with VV:751

1

FIG.2. Radiosequence analysis of -9-kDa carboxyl-termi-

nal fragment isolated from VV:751-infected CV-1cell lysates. Panel a, sequence of [35S]methionine-labeledfragment; panel b, sequence of [3H]lysine-labeled fragment; panel c, sequence of [3H] isoleucine-labeled fragment; panel d, partial sequence of 8-amyloid peptide illustrating secretase cleavage sites a t Lyd6 and Phelg (denoted by arrows). Gly 3.8 at cycles16 and 19 (Fig. 2a). When the experiment was repeated with protein labeled with [3H]lysine or [3H]isoleucine, 3Hradioactive signals were seen at cycles 9 and 12 from the lysine labeling (Fig. 2b) and at cycles 12, 13, 15, and 16 for isoleucine labeling (Fig.2c). When these data are aligned with the knownsequence of p-APP, a uniform fit for all radiolabeled amino acids indicates that the-9-kDa fragment

B

Sequence of fragment Sequence A of fragment

Cycle

Aminoacid

1st run"2ndrunAminoacid

Pml

Leu 6.3 Phe Ala 5.5 Val Phe 4.8 Glu 5.220 Asp Phe 25 3.5 5 Ala 16 Val 3.5 166 Glu 18 2.9 Ser 7 Asp 128 Val 2.6 Asn 9 Gly 10 17 Lys 16 Gly 10 Ser (1.25) Ala 13 11 Asn Ile 7 12 Lys a Power failure in the first three cycles. 1 2 3 4

1st run'2ndrun

-

PWl

17 18

(2.44) 16 11 14

4.5 3.8 3.5 3.2 2.9 (1.2) 2.4

@-APPSecretase Cleavage Sites

630

series of mutants with deletions of up to 6 amino acid residues amino-terminal to and including Lys" as well as 6 residues immediately following Lyd6 were constructed and tested for production of soluble 8-APP695. All of these deletion mutations showed secretion of precursor protein into the culture medium (data not shown). Two of the deletion mutants lacking 4and6 residues immediately adjacent to the carboxyl-terminal side of L y P (KM-3 and KM-4, respectively) were investigated with respect to thecleavage site(s) used for secretion. These mutants were selected because they each retain Lys" but also harbor significant deletions and might, therefore, reveal specific alterations in secretion site(s). To increase the expression efficiency and production of -9-kDa fragments, DNAs for these two mutants were individually engineered into vaccinia viruses. CV-1 cells were infected with the recombinant vaccinia viruses, metabolically labeled with [?'3]methionine/cysteine, and cell lysates prepared. Equivalent amounts of labeled lysate from each infection were immunoprecipitated with 369Ab and analyzed by electrophoresis on SDS-polyacrylamide gels (Fig. 1,D andE ) . The carboxyl-terminal fragments from the KM-3 and KM-4 cell lysates gave slightly higher molecular mass bands in the -9-10 kDa range than theirwild type counterpart, suggesting that the secretory cleavages in these deletion mutants occurred amino-terminal to Lys". Sequencing analyses were next performed to identify the exact cleavage sites used to release soluble @-APPfrom the KM-3 and KM-4 mutants. The -9-kDa fragments present in [35S] methionine metabolically labeled KM-3 andKM-4 cell lysates were isolated and sequenced. 35Ssignals were found at cycles 18 and 20 for the KM-3 mutant,suggesting a two-site cleavage

after Hid3 and Glu" within the @-amyloiddomain (Fig. 4, a and c ) . Identical cleavage sites were found upon analysis of the KM-4 mutant protein harboring a larger 6-amino acid deletion. 35Sradiosignals were observed at cycles 16 and 18 indicating a double cleavage, again at Hid3and Glu" (Fig. 4, b and c ) . The secretion sites for KM-4 are located 9 and 11 residues amino-terminal to the putative point of insertion into the membrane as surmised from the hydrophobic and hydrophilic nature of the sequence in this region. In contrast, the Lys" site is located 12 residues from the transmembrane domain. Secretory Cleavage Sites in Down's Syndrome FibroblustsTo compare the secretory cleavage sites identified in human P-APP synthesized by recombinant methods with the secretion of endogenous P-APP in a human cell line, we obtained a fibroblast cell line derived from the skin of an individual with Down's syndrome. The DSfibroblasts were metabolically labeled with [35S]methionine/cysteine,cell lysates prepared, immunoprecipitated with 369Ab, and analyzed by SDS-polyacrylamide gel electrophoresis. The pattern of carboxyl-terminal fragments obtained was very similar to that seen produced in CV-1 fibroblasts (Fig. 1F).When the -9-kDa fragment from DS fibroblasts was radiosequenced, a heterogeneous collection of cleavage sites was found. 35Sradiosignals were observed at cycles 10, 13, and 16, indicative of multiple cleavages (Fig. 5). Based on these data, it can be concluded that one cleavage site occurred after Phel9 of the @-amyloiddomain, with additional sites afterG1u22and Gly25. In contrast to the -9-kDa fragment generated by VV751 infected CV-1 cells, no radiosignal was found at cycle19, which is expected if cleavage occurred at Lys". Hence, the proteolytic cleavage sites used for P-APP751 secretion in DS fibroblasts are heterogeneous and different from those found in the CV-1 cell system. DISCUSSION

One potential mechanism to explain the excessive production of P-amyloid peptide in thebrains of Alzheimer's disease victims is that the peptide is generated upon secretion of @APP. Therefore, an understanding of the proteolytic cleavage process used to release soluble B-APP is important in elucidating the role of secretion in P-amyloidogenesis. A number 0 5 10 15 20 25 30 of investigators have studied @-APPsecretion. Esch et al. (8) first defined the major secretase cleavage site as Lys" of the cycle FIG. 3. Radiosequence analysis of [s5S]methionine-labeled &amyloid domain based on biochemical analyses of both recombinant 8-APP695 and P-APP751 proteins synthesized -B-kDa carboxyl-terminalfragmentisolated fromchloroin human 293 cells. Subsequently, similar results have been quine-treatedVV761-infected CV-1 celllysates. TABLE I1 Summary of /j"APPprocessingmutants @-Amyloiddomain

. . . .1 16 Membrane . . . .DAEFRHDSGYEVHHQKLVFFAEDVGSNK§.. . L A

R E

Q AA

A AA AA AA AAAA AA

A AA AAAA AAAAAA

Mutants

WT K16L K16A K16R K16E Kl6Q K16A/L17A AK16 AQ15, K16 AH13, H14, Q15, K16 AEll,H14 V12, H13, Q15, K16 AL17 AL17, V18 (KM-3)AL17, F20V18, F19, AL17, V18, F20,F19, A21, (KM-4) E22

P-APP Secretase Cleavage Sites

631

ance liquid chromatography retentiontimeasasynthetic peptide standard designed to correspond to cleavage at Lys" was analyzed. In the original study by Esch et al. (8), as well as those of Anderson et al. (19), and Ramabhadranet al. (21), molar yields of /i?-APPfragments characterized ranged from 2 to 12% depending on the study. Hence, sites other than the major Lys" cleavage site may not have been detected. In fact, this caveat was noted by Ramabhadran et al. (21). Experiments in which the Lys" cleavage sitehas been mutated or extensively deleted revealed little negative effect on the secretion of p-APP (15-17). The secretase cleavage site used to secrete the various p-APP mutants was not identified except for one deletion mutant generated by Maruyama et al. (17). Based on the position of the cleavage site found for this one large deletion mutant spanning the Lys" KM4 16 site, itwas concluded that secretase is not a sequence-specific proteinase but is distance-specific, i.e. the site cleaved in pAPP is solely determined by its distance from the membrane surface. Since most proteinases demonstrateamino acid specificity, we investigated the possibility that a collection of proteinases, with different specificities, may be responsible for the secretion of p-APP. With this scenario, the primary cleavage at Lys" would be produced by a prevelant proteinase with trypsin-like activity in thecollection of secretases. Deletion ofLysl' would then allow another proteinase in the 0 s 10 1s 2 0 2 5 30 collection with a different specificity to cleave and secrete pCycle APP. If a collection of proteinases comprises the secretase C) activity, then itwould bepredicted that otherp-APP cleavage sites should exist. In addition, it is possible that thesecretase l6 YEVHHQK{(LVFF)AE)DVGSNK cleavage area represents a structurally exposed region of pAPP which is vulnerable to proteolytic cleavage. The apparent AAAA KM3 preference for secretase to cleave at a set distance from the m 4 A A A AA A FIG.4. Radiosequence analysis of [s6S]methionine-labeled membrane surface could be accounted for by a susceptible -9-kDa carboxyl-terminal fragments produced by deletion region of amino acid sequence in p-APP. To begin to refine our understanding of the p-APP secretase mutants of @-APP.Panel a, KM-3 -9-kDa fragment analysis;panel b, KM-4 -9-kDa fragment analysis; panel c, partial sequence of 8- cleavage process, we have closely examined the cleavage site amyloid peptide with secretase cleavage sites of KM mutants indi- by conducting radiolabel and direct amino-terminal sequence cated by arrows. KM-3 and KM-4 deletions are also indicated on the analyses of the -9-kDa carboxyl-terminal product of p-APP sequence. The -9-kDa fragments were generated by infecting CV-1 secretion. Recombinant p-APP751 synthesized in CV-1 fibrocells with VVKM-3 or VVKM-4. blasts was one source of -9-kDa cleavage product characterized. Radiosequence data using three different amino acids to 13 label the -9-kDa fragment metabolically were consistent with two sites of cleavage, one at Lys" and anotherat Phelg.Direct amino-terminal sequence of the fragment confirmed the radiosequence results. These same cleavage sites were obtained when the -9-kDa fragment was produced in the presence of chloroquine, a drug that suppresses protein degradation by acidifying intracellular proteolytic compartments. These results suggest that there is heterogeneity in secretase cleavage which is not caused by contributing degradative processes. Furthermore, the nature of the residues in the P1 positions would suggest that proteinases of different specificities are O l 0 5 10 1s 20 25 30 operating in the secretion event. This conclusion is further cycle substantiated by analysis of p-APP mutants made at and FIG.5. Radiosequence analysis of [s6S]methionine-labeled around the primary cleavage site, Lys" of (3-amyloid. -9-kDa fragment from DS fibroblasts. Thirteen different mutations of p-APP, including five point mutations at Lys" and eight deletional mutations removing obtained for human p-APP expressed in Chinese hamster 1-6 residues at and adjacent to Lys", were constructed and ovary cells (16, 18), COS-1 cells (15-17), SF9 insect cells (20, tested for the ability to secrete p-APP. All mutants were 21), and native p-APP synthesized in PC-12 neuronal cells found to produce soluble 6-APP. This result is in agreement (19) and Alzheimer's disease brain (22). However, in all of the with Sisodia (15), Sahasrabudhe et al. (16), andMaruyama et studies, methodology and/or yields of protein may have only al. (17), who also found that mutations in this region of 6permitted the detection of a major cleavage site by secretase. APP would still permit secretion of p-APP. We defined the For example, in the study conducted by Wang et al. (18)in cleavage sites for two deletion mutants with preserved Lys". which the carboxyl terminus of soluble B-APP wascharacter- Metabolically labeled -9-kDa fragments were purified from ized, only the CNBr product with an identical high perform- cells expressing each mutant and radiosequenced. The results 18

...

++

KM3

...

632

B-APP Secretase Cleavage Sites

indicate that for both mutants, cleavage after Lys16is ignored in favor of cleavage after other amino acids, namely, His13 and Glu" of the p-amyloid domain. These data again suggest heterogeneity in secretase cleavage. Exopeptidase activity does not appear to be responsible for generating this heterogeneity. The cleavage sites for wild type p-APP751 are separated by 3 amino acids, whereas the sites found for both pAPP751 mutants are separatedby 2 amino acids; hence, it is difficult to envision selective activity by tripeptidyl and dipeptidyl aminopeptidases in the same cell background. The absence of cleavage at the retained Lys16 may reflect conformational alterations imposed by the deleted residues or may be a result of the loss of residues required for proteinase specificity resulting from deletion of the P1' and P2' positions adjacent to the P1 Lys". The newcleavage sites used to secrete the mutated p-APPs are located up-stream of L y P similar to thatobserved by Maruyama et al. (17), who studied a different deletion mutant of p-APP which also cleaved after a glutamate residue. Although the two mutants differed in the lengths of residues deleted (4 and 6 amino acids), the cleavage sites were identical. The use of the same cleavage site with both mutants suggests cleavage by an alternate proteinase(s) with a defined specificity versus cleavage by a distance-specific proteinase. Further evidence for heterogeneity of secretase cleavage comes from our analysis of native p-APP synthesized by DS fibroblasts. Cleavage was not found a t Lys16 but rather at Phe", GluZZ,and G1yz6 ofthe @-amyloiddomain. Since it has been demonstrated that tripeptidyl and dipeptidyl aminopeptidase activity is unalteredin Alzheimer's disease (27), it seems unlikely that the multiple cleavage sites found in the DS fibroblasts result from tripeptidyl aminopeptidase action. The distances from these sites to the membrane surface are 9, 6, and 3 residues. These data deviate from the suggested specific distance of9-11 residues. Whetherthis cleavage pattern is unique to Down's syndrome, Alzheimer's disease, or to cell type variation requires further investigation. Heterogeneity of secretase action has also been recently reported by Seubert et al. (28) who employed specific antibodies to identify alternative cleavage sites used in p-APP release. In addition, our results describing secretase heterogeneity can explain the dataof Busciglio et al. (29) who characterized the 3-kDa product generated by subsequent processing of the -9kDa fragment. Busciglio et al. (29) reported a heterogeneous amino-terminal sequence for the 3-kDa peptide reflecting cleavages at Lys16,Phelg, andValz4,which we believe is caused by the action of multiple secretory proteinases. In summary, this study demonstrates that the secretion of p-APP is likely to involve a collection of proteinases that have been heretofore referred to as secretase. Cleavage ofj3APP also appears to be heterogeneous, occurring at several sites in a region of the precursor near the membrane surface. Although Lys" of the @-amyloiddomain seems to be a pre-

ferred cleavage site for most cell systems studied, alternative sites may be used in differentcellular backgrounds. Acknowledgments-We thank Drs. Sam Gandy and Paul Greengard for 369Ab, Daiichi Pharmaceutical Co. for support with mutant production, and Marion Merrell Dow Inc. for support of all other experiments. REFERENCES 1. Kang, J., Lemaire, H.-G., Unterbeck, A,, Salbaum., J. M., Masters C. L.,

Grzeschik, K. H., Multhaup, G., Beyreutber, K., and Muller-Hill, B.

(1987) Nature 3 2 6 , 733-736 2. Ponte, P., Gonzalez-DeWhitt, P., Schillin

3. 4. 5. 6. 7. 8.

,J , Miller, J., Hsu, D., Greenberg, B., Davis, K., Wallace, W., Liebegurg, I., Fuller, F., and Cordell, B. (1988) Nature 331,525-527 Kitaguchi, N., Takahashi, Y., Tokushima, Y., Shiojiri, S., and Ito, H. (1988) Nature 331,530-532 Tanzi, R. E. McClatchey, A. I., Lamperti,E. D., Villa-Komaroff, L., Gusella, J. k., and Neve, R. L. (1988) Nature 331,528-530 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 Weidemann, A., Konig, G., Bunke, D., Fiacher, P., Salbaum, J. M., Masters, C. L., and Beyreuther, K. (1989) Cell 67,115-126 Gandy, S, E., Czernik A. J., and Greengard, P. (1988) Proc. Natl. Acad. Sci. U. S. A. 8 6 , 6 2 l b 2 2 1 Esch, F.S., Keim, P. S. Beattie, E. C., Blacher, R.W.,Culwell, A. R., Oltersdorf, T., McClurk, D., and Ward, P. J. (1990) Science 248, 1122-

1124 9. Sisodia, S. S . , Koo, E. H., Beyreuther, K., Unterbeck, A., and Price, D. L. (1990) Sclence 248,492-495 10. Estus, S., Golde, T. E., Kunishita, T., Blades, D., Lowery, D.,Eisen, M.,

Usiak, M.,Qu, X., Tabira, T., Greenberg, B. D., and Younkin, S. G. (1992) Science 265, 726-728 11. Palmert, M. R., Poldisn , M. B., Witker, D. S., Oltersdorf, T., Younkin, L. H., Selkoe, D. J., and{ounkin, S. G. (1989) Proc. Natl. Acad. Sci. U. S. A.

86,6338-6342 12. Overly, C.C., Fritz, L. C., Lieberburg, I., and McConlogue, L. (1991) Biochem. Bio h s Res Commun. 181,513-519 sh:oi, J., Anderson, J. P., Pappolla, M. A., and Robakis, 13. Sambamurti, N. K. (1992) J. Neurosci. Res. 33,319-329 14. Haass, C., Koo, E. H., Mellon, A., Hung, A.Y., and Selkoe, D. J. (1992) Nature 367,500-503 15. Sisodia, S. S. (1992) Proc. Natl. Acad. Sci. U. S. A. 89,6075-6079 16. Sahasrabudhe, S. R., Spru M.A., Muenkel, H. A., Blume, A. J., Vitek, M. P., and Jacobsen, J. p(1992) J. Biol. Chem. 267,25602-25608 17. Maruyama, K., Kametani, F., Usami, M., Yamao-Harigaya, W., and Tanaka, K. (1991) Biochem. Biophys. Res. Commura. 179,1670-1676 18. Wang, R., Meschia, J. F., Cotter, R. J., and Siscdia, S. S. (1991) J. Biol. Chem. 2 6 6 , 16960-16964 19. Anderson, J. P., Esch, F. S., Keim, P. S., Sambamurti, K., Lieberburg, I., and Robakis, N. K. (1991) Neurosci. Lett. 128,126-128 20. Lowery, D. E., Pasternack, J. M., Gonzalez-DeWbitt, P. A,, Zurcher-Nee1

x

6:

H., Tomich, C . 3 . C., Altman, R. A., Fairbanks, M. B., Heinrikson, L., Younkin, S. G., and Greenberg, B.D. (1991) J. Biol. Chem. 266,

19842-19850 21. Ramabhadran,

T.V., Gandy, S. E., Gbiso, J., Czernik, A. J., Ferris, D., Bhasin, R., Goldgaber, D., Frangione, B., and Greengard, P. (1993) J. Biol. Chem. 268,2009-2012 22. Pastemack, J. M., Palmert, M.R., Usiak, M., Wang, R., Zurcher-Neely, H., Gonzalez-DeWhitt, P. A., Fairbanks, M. B., Cheun , T , Blades, D., Heinrikson, R. L., Greenberg, B. D., Cotter, R. J., an$ Younkin, S. G. (1992) Biochemistry 31.10936-10940 23. Wolf, D., Quon, D., Wang, Y., and Cordell, B. (1990) EMBO J. 9, 20792084 24. Mu hy, G. M., Tinklenlxr, J. R., Eng, L. F., Cordell, B., Wan , Y , Ellis $G., and Meissner, L. (1990) Prog. Neuro-psychopharmacof 14, 3091 317 25. Gabuzda, D., Olshevsky, U., Bertani, P., Haseltine, W. A., and Sodroski, J. (1991) J. Acquired Immune Defk. Syndr. 4,34-40 26. Caporaso, G. L., Gandy, S. E., Buxbaum, J. D., and Greengard, P. (1992) Proc. Natl. Acad. Scr. U.S. A. 89,2252-2256 27. Mantle, D., and Perry, E. K. (1990) J. Neurol. Sci. 98, 13-20 28. Seubert, P., Oltersdorf, T., Lee, M. G., Barbour, R., Blomquist, C., Davis,

D. L.. Brvant. K.. Fritz. L. C.. Galasko. D.. Thal. L. J.. Lieberbure. I.. ~

~

~~

Y

l

I

and ScheAk:D.-B: ( 1 9 9 3 ) Nature 361,260-263 ' ' 29. Busciglio, J., Gabuzda, D. H., Matsudaira, P., and Yankner, B. A. (1993) Proc. Natl. Acad. Sci. U. S . A. 90,2092-2096