To define the APP cleavage site, we constructed a. Chinese hamster ovary cell line, which constitutively overexpresses human APP-770, and analyzed the.
THEJOURNAL (C 1991 by
OF BIOLOClCAL CHEMISTRY The American Society for Biochemistry and Molecular Biology, Inc.
Vol. 266, No. 25, Issue of September 5, pp. 16960-16964, 1991 Printed in U.S.A.
Secretion of the PIA4 Amyloid PrecursorProtein IDENTIFICATION OF ACLEAVAGE
SITEINCULTURED
MAMMALIAN CELLS* (Received for publication, March 15, 1991)
Rong WangS,James F. MeschiaV, Robert J. Cotter$, and Sangram S.SisodiajTIl From the Departments of $Pharmacology and Molecular Science and §Pathology and the VNeuropathology Laboratory, T h e Johns Hopkins University School of Medicine, Baltimore, Maryland21205-2181
Alzheimer’s disease, a progressive neurodegenerative disorder, affects >lo%of the population of individuals >65 years of age. A principal neuropathological feature of this disease is the senileplaque, a fibrillar extracellular deposit primarily composed of a -4-kDa peptide, B/A4, derived from the amyloid precursor protein (APP). Studiesin cultured cells have documented that APP matures through a constitutive secretory pathway and is cleaved at or near the cell surface to release a large ectodomain into the extracellular space. To define the APP cleavagesite, we constructed a Chinese hamster ovary cell line, which constitutively overexpresses human APP-770, and analyzed the COOH termini of secreted APP-770-relatedmolecules. Using plasma desorption mass spectrometry and chemical microsequencing, we document that an APP cleavage site in Chinese hamster ovary cells leading to secretion occurs immediately COOH-terminal to lysine residue 687, which lies adjacent to the hydrophobic membrane-spanning domain.
that APPs have a short intracellular half-life and are posttranslationally modified both by N - and O-linked carbohydrate structures and by tyrosine sulfation (13). Full-length molecules are detected on thecell surface, and proteinslacking the transmembrane and cytoplasmic domains are secreted into the medium. Soluble APP isoforms have been identified as protease nexin I1 (14,15). Similar events presumably occur in uiuo, based on the demonstration of COOH-terminally truncated APP molecules in the cerebrospinalfluid (13, 16). Recently, molecular biological studies identified a domain of APP thatwas necessary and sufficient for this constitutive cleavage event and provided evidence that this constitutive cleavage occurs in the extracellular domain, which lies just NHz-terminal to the APP transmembrane domain. Significantly, this cleavage is within the PIA4 peptide (17). Subsequently, Esch et al. (18) used a different approach to chemically identify unique COOH- and NHz-terminal residues of the secreted and membrane-retained fragments, respectively. However, the precise cleavage site was indeterminate because a lysine residue, which normally resides betweenthe identified termini of secreted and membrane-retained fragments, was not detected in this analysis (18). To definethe precise APP cleavage sitethat leads to A classic neuropathological feature of Alzheimer’s disease secretion, we have now generated a Chinese hamster ovary is the presence of numerous senile plaques in the cerebral cortices of affected individuals (1-3). The severity of these (CHO) cell line that overexpresses the human APP-770. Solintraneuronal, extracellular lesions in certain regions of the uble APP-770(SAPP-770), purifiedfrom theconditioned brain correlates with thepresence of dementia (4). The prin- medium of these cells, was chemicallyfragmented with cyancipal component of amyloid is a -4-kDa peptide (5), termed ogen bromide (CNBr). Peptides that contain COOH-terminal ,8/A4, derived from the amyloid precursor protein (APP)’ (6, residues of SAPP-770 were identified by plasma desorption verified by chemical 7). Four APP messenger ribonucleic acids (mRNA), derived mass spectrometry (PDMS) (19) and microsequencing. In this in uitro system, the COOH termini by alternative splicing of APP pre-mRNA transcripts, have been described that encode integral membrane glycoproteins of SAPP-770 molecules are heterogenous and differ by the of 695,714,751, and770 amino acids (7-11). The 42-43 amino presence or absence of a COOH-terminal lysine residue. acid PIA4 peptide includes 14 amino acids of the transmemMATERIALS AND METHODS brane domain and28 amino acids of the adjacent extracellular Cell Culture and Transfection-CHO cells, obtained from Dr. Se domain of APP (5,12).However, the biochemical events Jin Li (Carnegie Institute of Washington,Baltimore, MD), were responsiblefor the generation of PIA4 from APP are not maintained in Delbucco’s modified Eagle’s medium, containing 10% clear. Studies performed in cultured cells have documented fetal calf serum. Plasmidp770 contains sequences that encode human
* This work was supported by National Science Foundation Grants DMB 86-10589, United States Public Health Service Grants AG 05146 and NS 20471, The Metropolitan Life Foundation, and The Robert L. and Clara G. Patterson Trust. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked “uduertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. 11 To whom correspondence should be addressed: Neuropathology Laboratory, The Johns Hopkins University School of Medicine, 600 N. Wolfe St., 509 Pathology Bldg., Baltimore, MD 21205-2181. ’ The abbreviations used are: APP, amyloid precursorprotein; C H O , Chinese hamster ovary; SAPP, soluble amyloid precursor protein; PDMS, plasma desorption mass spectra; SDS-PAGE, sodium dodecyl sulfate-polyacrylamide gel electrophoresis; HPLC, high performance liquid chromatography.
APP-770 and includes the entire 5”untranslated region and -280 bases of the 3’-untranslated sequence. The APP-770 complementary deoxyribonucleic acid (cDNA) was inserted downstream of the SV40 early promoter/enhancer region (17). To generate stable cell lines that overproduce human APP-770, CHO cells were cotransfected withplasmids p770 and pSV2neo (20) by the calcium-phosphate coprecipitation technique (21). After transfection, cells were subcultured in complete medium containing 0.4 mg/ml G418. The G418resistant colonies that appeared after -2 weeks were isolated and expanded in selective medium. For selection of APP-770-overexpressing cell lines, cytoplasmic ribonucleic acid (RNA) was prepared from -lo6 cells, andsteady-state levels of exogenous transcripts were determined by S1 nuclease mapping (22). A CHO cell line, 770A5, positive for human APP-770 transcripts, was further characterized by immmunoprecipitation of [”S]methionine-labeledAPP-770-related polypeptides from cells and conditioned media and was visual-
16960
Cleavage Site of Amyloid Precursor Protein ized following fractionation by SDS-PAGE, fluorographic enhancement, and autoradiography, as previously described (17). Purification of Secreted SAPP-770 and CNBr Cleavage-Approximately 5 x 10' cells were plated onto60 150-mm tissue culture dishes in complete medium (Dulbecco's modified Eagle's medium + 10% fetal calf serum + 0.2 mg/ml G418). The media were decanted 24 h later, and cells were incubated in a serum-free growth medium consisting of MCDB 302 medium (Sigma) supplemented with Pen-strep (105 units/liter), L-glutamine (292 mg/liter), L-proline (12 pg/liter), and 0.06% NaHCOa for 4-6 h. These media were then decanted and replaced with 18 ml of MCDB 302. Serum-free conditioned media were pooled 72 h later, cell debris was removed by centrifugation, and thesupernatant fraction was concentrated at 4 "C to15 ml by sequential ultrafiltrationunits (Amicon models 8400 and 8050) equipped with Amicon YM-10 membranes with a nominal molecular mass cutoff of 10 kDa. This material was further concentrated sequentially with Centricon Centriprep 10 and Centricon 10-spin concentrators with nominal molecular mass cutoffs of 10 kDa. This final concentrate was divided into three parts, and each part was loaded onto a Pharmacia LKB Biotechnology lnc. Mono-Q HR5/5 anionexchange fast protein liquid chromatography column equilibrated in a buffer system of 2 mM 2-mercaptoethanol and 20 mM Tris-HC1, pH 7.4 (Buffer A). The protein was eluted at 1ml/min with a programmed gradient thatentailed a 5-min 100% Buffer A wash, followed by a 60min linear gradient to 100% Buffer B (Buffer A plus 1 M NaCl). Polypeptides in selected fractions were analyzed following fractionation by SDS-PAGE and silver staining. APP-770-relatedpolypeptides were identified by immunoblottingwitha monoclonal antibody (22Cll) directed against an amino-terminalepitope of APP (13). Peakfractions containing APP-770-related polypeptides from three Mono-Q runs were pooled and concentrated by spin centrifugation with a Centricon-10 microconcentrator. The sample was desalted on an NAP-5 (Pharmacia) column, previously equibrated in 0.5 M Tris-HC1. pH 8.6. The proteinwas subsequently denatured with the addition of guanidine HCI to 100 mM and sulfhydryl groups reduced in the presence of 0.2 M dithiothreitol at 37 "C for 90 min in the dark. The reduced cysteines were alkylated by the addition of iodoacetamide to 0.25 M, while maintaining the pH at -8-8.6. This reaction was performed at room temperature for 30 min, quenched with excess P-mercaptoethanol, and thendesalted on a NAP-5 column preequilibrated with 50 mM ammoniumformate. The sample was lyophilized and then resuspended in 70% (v/v) trifluoroacetic acid. Approximately 10 mg of CNBr was added, and the mixture was incubatedovernight in the dark at room temperature. Following evaporation of volatile components with a stream of nitrogen, 0.5 ml of H,O was added, and the mixture was lyophilized to dryness. HPLC Fractionation of CNBr Peptides and PDMS AmlyskCNBr-derived SAPP-770 peptideswere dissolved in 60% (v/v) trifluoroacetic acid and applied to a VYDAC CIR reverse-phase HPLC column (5 pm, 4.6 X 250 mm) attached to Spectra-Physics model SP8800 HPLC system. Peptides were eluted from the column at a constant flow rate of 1.0 ml/min over a total time of3.5 h with a stepped linear elution gradient. The mobile phases were:A, 0.1% trifluoroacetic acid in water; B,0.1%trifluoroaceticacid in 85% acetonitrile; and C, 0.1% trifluoroacetic acid in acetonitrile. Initially, a linear gradient from 100% A to 70% A and 30% B was applied for 2 h. The percentage of B was then increased to 60% during 2-3 h, after which time a linear gradient to 100% C was applied during a 20-min period. Eluted peptides were detected by monitoring at 215 nm. To identify a region of the profile that contained peptides that include the COOH- terminus of SAPP-770, we fractionated a peptide, DAEFRHDSGYEVHHQK,a potentialSAPP-770 CNBr-derived COOH-terminal peptide (18). This peptide had a retention time of -66 min under these chromatographic conditions (not shown). The PDMS of selected peaks within this region was obtained on a BIO-ION Nordic (Uppsala,Sweden) BIN-1OK plasmadesorption time-of-flight mass spectrometer equippedwith a 10-pCi Californium252 ionization source and a P D P 11/73-based data system. One-fifth of each fraction was lyophilized, dissolved in 0.1% trifluoroaceticacid in water, and deposited on nitrocellulose-covered, Mylar-backed aluminum foil. Positive ion mass spectra were recorded with an accelerating potential of +18 kV for five million primary events. Negativeion mass spectra were recorded with an accelerating potential of -15 kV for five million primary events. The time resolution was 1 ns/ channel. H+ andNa' ions were used for positive spectra calibration, and H- and CN- ions were used for negative spectra calibration. A partial PDMS spectrum of fraction 66 revealed a major ion of m / z 1955.6 (Fig. 4B), anda partial PDMS spectrumof fraction 67 revealed
16961
Ezlrocellular
B. 687
677
697
FIG. 1. A , transmembrane orientation andposition of @/A4region of APP-770. Darkened region depicts signal peptide. Wavy lines indicate PIA4 region. CHO represents position of N-linked carbohydrate sites. The position of the membrane and the protease inhibitor domain are indicated. B, amino acid sequence NH2-terminal to APP transmembrane domain. The boxed area includes 28 residues of the amino terminus of P/A4. Arrowheads indicate termini identified by Esch et al. ( l a ) , resulting from a membrane-associated cleavage event. Arrow indicates primary site of cleavage identified in the present study. Numbering is after Kang et al. (7). a major ion of m / t 1826.9 (Fig. 4C). Negative-ion PDMS spectra show a -2-atomic mass unit decrease for both peaks (data notshown) and confirmed that theions correspondedto the protonated molecules in thepositive ion spectra. Amino Acid Sequencing-The remaining samples from each fraction that contained peptides of appropriate molecular weight, assigned by PDMS analysis, were lyophilized and then subjected to microsequencing on an Applied Biosystems model 470 sequenator. Phenylthiohydantoins were separated on a 21 X 220-mm CIRcolumn, and yields at each cycle were determined by comparison with phenylthiohydantoin standardsapplied duringthe precycling of the sequencing system. RESULTS
Expression and Purification of SAPP-770-related Polypeptides-Esch et al. (18) had previously documented that the constitutive cleavage of APP results in a secreted molecule with glutamine 686 at its COOH terminus and a membraneretained fragment withleucine 688 at itsNH' terminus. These results indicated that primary cleavage could occur on either side of lysine 687 (see schematic of the structure and transmembrane orientation of APP-770 in Fig. 1A and the amino acid sequence NH2-terminal to the APPtransmembrane domain in Fig. 1B). To identify the precise cleavage site, we purified the secreted form of human APP-770 from conditioned media of CHO cells that constitutively overexpress a human APP-770 polypeptide. We elected to characterize the COOH terminus of the secreted fragment instead of the NH, terminus of the membrane-retained fragment because pulsechase experiments revealed that the COOH-terminal fragment has an extremely short half-life in CHO cells.' Purification of the membrane-retained fragment would pose a considerable challenge. Moreover, the secreted molecule has an indefinite half-life in conditioned media. CHO cells were chosen for these experiments for the following reasons. CHO cells express low levels of APP-770 and APP-751 and secrete COOH-terminal deleted APP-related polypeptides,2the SV40 early promoterregion is transcriptionally active in thesecells, and CHO cells can be cultured in defined a serum-free medium for up to 72 h (23). We screened 10 independent CHO cell lines for expression of human APP-770 transcripts using S1 nuclease protection analysis of steady-state mRNA (not shown). A CHO cell line, which we refer to as 770A5,was subsequently selected, which constitutively overexpressed the human APP-770 cDNA. APP-770 expression in line 770A5 was assayed by culturing this cell line in the presence of ["SI methionine, and radiolabeled APP was immunoprecipitated from the cell and conditioned medium with a polyclonal antibody raised against bacterially synthesized APP-695 (13). S. S. Sisodia, unpublished observation.
Cleavage Site of Amyloid Precursor Protein
16962
Immunoprecipitates were visualized following fractionation by SDS-PAGE, fluorographic enhancement, and autoradiography. Line 770A5 was found to express significant amounts of APP-770 and to secrete APP-770-related molecules (compare untransfected cells (Fig. 2, lanes 1 and 2 ) with products of770A5 cells (Fig. 2, lanes 3 and 4 ) ) . A major, -110-kDa polypeptide (SAPP-770) is secreted into theconditioned medium by 770A5 cells (Fig.2, lane 6)but is notsecreted by the parental CHO line (Fig. 2, lane 5 ) . To purify SAPP-770 from the culture medium of line 770A5, we grew these cells in a defined, serum-free medium. The use of serum-free media substantially decreased contaminating polypeptides. Next, the mediarecoveredfollowing growth for 72 h and were concentrated and then chromatographed on an anion-exchange column (Mono-&). Individual fractions (Fig. 3A) were examined by SDS-PAGE and silver staining. The putative SAPP-770 polypeptide was the major constituent of the prominent peak eluting between 40 and 42 min (Fig. 3B). This polypeptide was APP-related, as was demonstrated by immunoblotting parallel samples with monoclonal antibody 22Cll that recognizes an amino-terminal epitope of APP (Fig. 3C). By comparison with the staining intensity of known amounts of phosphorylase b loaded in adjacent lanes of the SDS-PAGE, we calculated that -30 pg of SAPP-770 were recovered from1liter of medium. To determine the precise cleavage site of SAPP-770 from its precursor, we chemically fragmented -300 pg of SAPP-770 with CNBr and fractionated the resulting peptides by reversephase HPLC. A portion of the reverse-phase HPLC profile of peptides generated following CNBr fragmentation of SAPP770 is shown in Fig. 4A.This region of the chromatogram was selected for analysis because a synthetic 16-amino acid peptide, DAEFRHDSGYEVHHQK, had a retention time of -66 min under identical chromatographic conditions (not shown). This synthetic peptide corresponds to the predicted CNBrgenerated peptide of SAPP-770 if cleavage during secretion were to occur COOH-terminal to lysine 687. To confirm that 1 2 3 4
5 6
I
d
ii L0
0.0
-
10 20 30 40 50 80 70 80 90
FRACTION NUMBER
Ag Stain
Blot
FIG. 3. Purification of SAPP-770 from conditioned media of 770A5 cells. A , Mono-Q fastprotein liquid chromatography profile of concentrated conditioned media of770A5 cells. Twenty 150-mm plates containingnearly confluent monolayers of 770A5cells were maintained in a serum-free media for 72 h. Conditioned media was concentrated, and polypeptides were subsequently fractionated by anion-exchange fast protein liquid chromatography. The bar above the peak between fractions 40-42 represents SAPP-770 polypeptides. B, silver stain of peak fractions 40-42. Twenty pl of each fraction were separated on an 8-16% gradient SDS-PAGE, fixed, and then stained with silver. The arrow indicates -110-kDa SAPP-770 in peak fractions. C, immunoblot of peak fractions 40-42. Twenty pi of each fraction were separated by SDS-PAGE and thentransferred to nitrocellulose membranes. Following blocking in 5%nonfat dry milk, the membrane was incubated with an APP-specific monoclonal antibody, 22Cll. Bound antibody was detected with affinity purified secondary goat antimouse I g G (Cappell), and immune complexeswere visualized by peroxidase diaminobenzidine enhancement.
an authentic CNBr-generated COOH-terminal peptide of SAPP-770 fractionated in this region of the chromatogram, we obtained a mass spectrum of each peak by PDMS. This initial PDMS survey revealed that fractions 66 and 67 of the chromatogram (Fig. 4A) contained peptides with molecular weights appropriate for a CNBr-generated COOH-terminal peptide of SAPP-770. Fig. 4B depicts the PDMS analysis of fraction 66 showing a strong positive ion peak of mlz 1955.6 and a weaker positive ion peak of m/z 1827.1. Similarly, Fig. 4C depicts the PDMS analysis of fraction 67 showing peaks identical to fraction 66 but with reversed signal strengths. These two ions of1955.6 and 1826.9 [M+H]+, formed with the addition of a positively charged proton to theappropriate c s c 8 cno rn peptide molecule, correspond to the molecularweights of Imnunappt. CNBr-generated peptides of SAPP-770 with the sequence FIG. 2. Overexpression of human APP-770 in CHO cells. CHO cell line 770A5, which constitutively expresses human DAEFRHDSGYEVHHQK (calculated m/z = 1955.6) and APP-770, was selected for analysis. Plates (60 mm), seeded with DAEFRHDSGYEVHHQ (calculated mlz = 1826.9), respec-1.5 X 10 cells, were labeled for 4 h with 150 pCi of [3sS]methionine. tively. Note that the predicted difference in hydrophobicity APP-related moleculeswere immunoprecipitated from cell pellets of the peptides is reflected by the respective retention on the and conditioned media and were visualized following fractionation by reverse-phase column. Two additional peaks appear in Fig. SDS-PAGE, fluorography, and autoradiography. Lanes I and 3 rep4C at mlz 1514.5 and mlz 1541.5. Although it is not certain resent immunoprecipitates from cell pellets of untransfected CHO and 770A5 cells, respectively. Lanes 2 and 4 represent immunoprecip- whether these represent minor species or mass spectral fragitates from conditioned media from untransfected CHO and 770A5 ments, we favor the latterexplanation, since these masses are cells, respectively. Lanes 5 and 6 represent the total labeled products consistent with fragmentation at theC-C bond between hisin conditioned media from untransfected CHO and 770A5 cells, tidine 684 and histidine 685 (calculated a13ion 1515.6 and y12 respectively. Arrow indicates SAPP-770 in 770A5-conditionedmedia. Although not apparent at thislevel of exposure, APP-751 and APP- ion 1513.6) and at the amide bond between these histidine 770 signals inuntransfected CHO cells can be detected at long residues (calculated b13 ion 1543.6). Finally, we confirmed the authenticity of peptides conexposures.'
$2
Cleavage Site of Amyloid Precursor Protein 0.02,
A
I
16963
I
5.0
0.0
42.0
Fract. 67 24.0
-1
8.0
0.0
2000
1
MI2
tained in fractions 66 and 67 by microsequencing. Fig. 5, A and B, depict the yields of prominent amino acids detected at each sequence cycle for peptides contained within fractions 66 and 67, respectively. These datarevealed that fractions 66 and 67 did contain peptides with amino acid sequences that matched the APP sequence. Furthermore, by comparison with the amino acid sequence presented in Fig. lB, it is clear that the SAPP-770 product in fraction66 corresponds to a peptide that initiates at aspartic acid 672 and terminates at lysine 687. These data areconsistent with the PDMS assignment of the peptide molecular weight. Similarly, in fraction 67, the sequence indicates a peptide initiating at aspartic acid 672 and terminating at glutamine 686, confirmed, again, by the prediction of the PDMS analysis. These data provide compelling proof that a primarysite for the membrane-associated cleavage of APP-770 is immediately NHz-terminal toleucine 688, thereby generating a secreted molecule (i.e. SAPP-770) with a COOH-terminal lysine residue. Whether a postcleavage event involving exopeptidase removal of lysine 687 or a bona fide proteolytic event between lysine 687 and glutamine 686 gives rise to secreted molecules with a COOH-terminal glutamine residue (as predominates in fraction 67), is presently uncertain.
3
4
5
6
7
8
9
1011121314151
SEQUOVCNG CYCLE
c FIG. 4. Identification of COOH-terminal CNBr-derived SAPP-770 peptides. A , reverse-phase HPLC of CNBr peptides of SAPP-770. Fractions 40 and 41 from 10 independent Mono-Q preparations were pooled, concentrated, and reacted with CNBr. Resultant peptides were fractionated by reverse-phase HPLC. A portion of the HPLC profile that represents peptides that have similar retention to asynthetic peptide homologous to a predicted CNBr-derived COOH-terminal peptide of SAPP-770 is shown. The arrow indicates peptide(s) with a retention of -66 min. B, PDMS analysis of fraction 66 from the chromatogram shown in Fig. 3A. A major molecular ion of1955.6 is apparent. C, PDMS analysis of fraction 67 from the chromatogram shown in Fig. 3A. A major molecular ion of 1826.9 is apparent.
2
Deduced sequence:
OAEFRHDSGYEVHHaK]
FIG. 5. Amino acid sequence of COOH-terminal CNBr-derived peptides of SAPP-77. A and B, amino acids of the highest molecular yields are shown at each sequence cycle of peptide(s) contained within fractions 66 and 67, respectively, of the HPLC profile. Shaded burs indicate residues that match the authentic APP sequence. Cross-hatched burs depict residues that do not match the predicted SAPP-770 sequence. C, deduced sequence of peptide(s) in fractions 66 and 67. DISCUSSION
This investigation corroborates earliermolecular biological studies localizing the constitutive cleavage of APP that occurs within the amyloidogenic @/A4region (17). Although Esch et al. (18) documented that this cleavage event resultsina secreted APP molecule containing a COOH-terminal glutamine residue (glutamine 686) and a membrane-retained fragment with an NH2-terminalleucine residue (leucine 688), the precise site of cleavagewas uncertain because lysine 687, normally present between these termini,was not detected and could have been removed from either of the identified termini by an exopeptidase. In the present analysis, we have characterized the COOH terminus of the secreted human APP-770 molecule in CHO cells. Analysis of the NH2terminus of the membrane-retained fragment was not attempted due to the extremely high turnover of that intermediate in CHO cells. Esch3 encounteredsimilar difficulties in their analysis of the membrane-retained fragment in human cells and recovered
