THE JOURNAL cw BIOWGICAI, CHEMISTRY G 1990 by The American Society for Biochemimy and M&xular Bioloa, Inc.

Vol. 26.5, No. 8, Iswe of March 15, pp. 4492-4497.1990 Printed in U.S.A.

The Alzheimer Amyloid Precursor Protein IDENTIFICATION OF A STABLE INTERMEDIATE IN THE BIOSYNTHETIC/DEGRADATIVE PATHWAY*

(Received for publication, July 27,1989)

Tilman Oltersdorf, Pamela J. Ward, Thomas Henriksson$, Eric C. Beattie, Rachael Neve@l, Ivan Lieberburg, and Lawrence C. Fritz From Athena Neurosciences Inc., South San Fran&q California 94080 and the $Division of Genetics, The Children’s Hospital, and Program in Neurosciences, Harvard Medical School, Boston. Massachusetts 0.2115

The amyloid forming @-peptide of Alzheimer’s dis- ease is synthesized as part of a larger integral mem- brane precursor protein (j!?APP) of which three alter- natively spliced versions of 695, ‘75I, and 770 amino acids have been described. A fourth /?APP form of 563 amino acids does not contain the @-peptide region. Re- cent experiments using transient expression in HeLa cells (Weidemann, A., Konig, G., Bunke, D., Fischer, P., Salbaum, J. M., Masters, C. L., and Beyreuther, K. (1989) Cell 57, 115-126) indicate that the @APP undergoes several posttranslational modifications in- cluding the cleavage and secretiou of a large portion of its extracellular domain. The nature and fate of the fragment that remains cell-associated following this cleavage has not heretofore been described. The me- tabolism of this fragment may have particular signifi- cance in Alzheimer’s disease since it must contain at least part of the @peptide. To study the metabolic fate of this fragment, we have established cell lines over- expressing the 6%5- and 751-amino acid versions of j3APP. P&e-chase studies show that this system is similar to the HeLa cell system in that both proteins are synthesized first as membrane-bound proteins of approximately 98 aud 106 kDa carrying asparagine- linked sugar side chains and are subsequently proc- essed into higher molecular mass forms by the attach- ment of sulfate, phosphate, and further sugar groups including sialic acid, adding approximately 20 kDa in apparent molecular mass. The mature form of &APP is cleaved and rapidly secreted, leaving an 11.5-kDa fragment with the transmembrane region aud the cy- toplasmic domain behind in the cell. This fragment is stable with a half-life of at least 4 h.

Alzheimer’s disease is the most common neurodegenerative disorder, affecting over 2 million people in the United States alone (for a review see Katzman, 1986). It is characterized by progressive dementia together with the presence of defining neuropathological features: extracellular deposits in the form of amyloid plaques and vascular amyloid, as well as intraceI- lular deposits in the form of neurofibrillary tangles. Amyloid plaques have also been found in Down’s syndrome and, in

* This work was supported in part by Eli Lilly and Co. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked “adver- &ement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

$ Current address: Cetus Corp., Emeryville, CA 94608. 11 Current address: Dept. of Psychobiology, University of Califor-

nia, Irvine, CA 92717.

smaller numbers, in normal aged human brain and in the brains of some aged mammals (for a review, see Selkoe, 1989). Ultrast~cturally, amyloid plaques and cerebrovascular amy- loid contain 6-lo-nm straight filaments, which are comprised of a 42-43-amino acid subunit, the so called @- or A4peptide (Glenner and Wang, 1984a, 1984b Masters et C& 1985). cDNA cloning has suggested that the @-peptide is synthesized as part of a much larger precursor molecule, the gene for which is located on chromosome 21 (Kang et &., 198% Tanzi et al., 1987; Goldgaber et ul., 198% Robakis et al., 1987). The sequence together with cell fractionation studies (Dyrks et ui., 19% Selkoe et al., 1988) suggest that the @APP is an integral membrane protein with the @-peptide spanning the border between the extracellular domain and the tr~smembrane region. To date, four ~ffe~nti~ly spliced versions of this /3- amyloid precursor molecule have been described comprising 695 amino acids @APP 695, Kang et ai., 1987), 751 amino acids (@APP 751, Ponte et 4, 1988 Tanzi et uZ., 1988), 770 amino acids (@APP 770, Kitaguchi et ul., 1988), and 563 amino acids (de Sauvage and Octave, 1989), respectively. The latter form does not contain the @-peptide region. The two larger forms carry a Kunitz-type protease inhibitor domain. The secreted versions of @APP with this protease inhibitor have been shown to be identical to protease nexin II (Oltersdorf et & 1989; Van Nostrand et ul., 1989). A key question regarding the molecular and cellular events that lead to amyloid depo- sition is how the membrane-associated @peptide is cleaved from its precursor and deposited extracellularly. It has re- cently been reported (Weidemann et ul., 1989) that @APP in HeLa cells is rapidly processed from a membrane-bound form to a secreted form that lacks the C terminus of the precursor. Weidemann et oi. (1989) indicate that the secreted forms derive from full length /3APP by cleavage in the extracellular domain, This would suggest the existence of a C-terminal membrane-associated fragment that remains behind in the cell following the cleavage event. Such a fragment has indeed been seen in extracts of human and rat brain (Selkoe et uZ., 1988). However, the relationship of this fragment to the pathway of @APP metabolism has not been investigated. Here, we describe the establishment of stably transfected 293 cells (transformed human embryonic kidney cells) expressing OAPP cDNAs. Using these cell lines we have analyzed the @APP metabolic pathway including synthesis, complex post- translational modifications, secretion, and ultimate formation of a long-lived, membrane-bound 11.5-kDa fragment.

EXPERIMENTAL PROCEDURES

Cell Lines-293 (transformed human embryonic kidney) and HS 683 (human glioma) cells were obtained from ATCC. Cell lines were grown as suggested by the supplier.

Expression of cDNAs for Human @APP in Human Cells-The

-4492

by guest on April 23, 2018

http://ww

w.jbc.org/

Dow

nloaded from

Alzheimer BAPP Metabolism 4493

establishment of 293 cells overexpressing flAPP695 from a cDNA driven by the human cytomegalovirus promoter/enhancer has been described earlier (Selkoe et ul., 1988). 293 cells overexpressing flAPP751 and HS 683 cells overexpressing pAPP695 were obtained using essentially identical methods. The @APP cDNAs were derived from an HL-60 promyelocytic leukemia cell line library (Tanzi et ul., 1988).

Expression of @APP Fragments in Escherichiu coli and Preparation of Antibodies-0APP fragments coding for amino acids 444-592 (re- striction sites BarnHI to gglI1) and 595-695 (restriction sites BglII to SpeI) were inserted in the BumHI restriction site of plasmid pExlOmer and the BarnHI and XbaI sites of plasmid pExl2mer, respectively (Seedorf er al., 1987). In order to introduce a stop codon in the polylinker, pExlOmer was modified by linearization at the XbaI site, fill-in with Klenow polymerase, and re-ligation of the obtained blunt ends. The natural stop codon of /3APP was used in the second construction. The resulting plasmids, designated pBX5 and DBX~. coded for fusion nroteins consisting of the loo-amino acid bacteriophage MS-2 polymerase N-terminal leader sequence followed by the respective @APP sequences. The fusion proteins were obtained by induction of the A PL promoter contained in these plasmids and partially purified by urea extraction as described (Seedorf er al., 1987). Rabbit antisera against the fusion proteins (anti-BX5 and anti-BX6) were affinity purified in two steps. First, to eliminate irrelevant antibodies, IgG fractions were prepared from the antisera and run over a column containing dialyzed urea extracts of bacteria expressing pExlOmer (and thus containing the MS-2 leader peptide) bound to cyanogen bromide activated-Sepharose (Pharmacia LKB Biotech- nologies Inc.). Second, the flow-through from the tirst step was run over another column containing the partially purified fusion protein, which had served as the antigen, coupled to activated Sepharose. Bound antibodies were eluted with 0.1 M glycine, pH 2.5, and were used for immunoprecipitation and Western blot experiments.

Metubolic Labeling and Immunoprecipitation-For metabolic label- ing, 5 x 106 to 1O’cells were incubated with 0.2 mCi of [“Slmethionine (1000 Ci/mmol, Amersham Corp.) in 2 ml of methionine-free Dulbec- co’s modified Eagles’s medium for 10 min. After two washing steps, cells were incubated for chase times from 20 min to 8 h in serum-free medium. Then the medium was collected and cells were lysed in TBS’ (10 mM Tris, pH 7.5, 150 mM NaCl) containing 1% Nonidet P-40, 5 mM EDTA, and 2 fig/ml leupeptin. Samples were preabsorbed with 15 mg of protein A-Sepharose (Pharmacia) for 1 h and immunopre- cipitated overnight at 4 OC in the presence of 5pg of specific antibody and 1.5 mg of protein A-Sepharose. Precipitates were washed four times in TBS, 0.1% Nonidet P-40, 5 mM EDTA, 2 pg/ml leupeptin, boiled in reducing SDS sample buffer, and loaded on standard Laem- mli SDS gels containing either 6.25% acrylamide or linear gradients of 7.5-20% acrylamide. After running, gels were fixed, incubated with AmplifyTM (Amersham Corp.), dried, and autoradiographed at -80 ‘C. For metabolic labeling with [35S]sulfate (25-40 Ci/mg, Amersham Corp.) or [‘*P]phosphate (1 mCi/ml, Amersham Corp.), 10’ cells in a 75~cm’ dish were grown in the oresence of 1 mCi of the label for 3 h in 3 ml of serum-&ee Dulbecco’s modified Eagle’s medium without additional sulfate or phosphate, respectively. Immunoprecipitation was done as described above.

1mmunoblotting-Cells were lysed in TBS, 1% Nonidet P-40, 5 mM EDTA, 2 fig/ml leupeptin. Protein from serum-free 48-h condi- tioned medium was concentrated by precipitation in 4 volumes of cold methanol. Proteins were separated on gels as described above. Transfer onto Immobilon P membranes (Millipore) was performed according to standard procedures. Filters were blocked in 10% skim milk and incubated with primary antibodies at a concentration of approximately 2 pg/ml. An alkaline phosphatase-coupled goat anti- rabbit antibodv (1:3000. Bio-Rad) served as second antibodv. Bound alkaline phosphatase activity was detected with nitro blue tetrazolium and 5-bromo:4-chloro-3-indolyl phosphate (Bio-Rad). Washing steps were carried out with TBS (10 mM Tris. DH 7.5. 150 mM NaCl) and 0.05% Tween 20.

Treatment with Endoglycosidase F and Neuraminidase-For diges- tion with endoglycosidase F (Boehringer Mannheim), immunoprecip- itates, obtained as described above, were washed in incubation buffer (50 mM sodium citrate, pH 5.0) and incubated overnight at 37 “C in 25 ~1 with 0.25 unit of enzyme. To control for nonspecific degradation, parallel samples were incubated in the absence of enzyme. For diges- tion with neuraminidase (Genzyme, Boston, MA), aliquots of 1%

‘The abbreviations used are: TBS, Tris-buffered saline; SDS; sodium dodecyl sulfate; HPLC, high pressure liquid chromatography.

Nonidet P-40 cell extracts or serum-free conditioned medium were precipitated with 4 volumes of cold methanol. Precipitates were resuspended in 100 ~1 of 0.1 M sodium phosphate, pH 6.5, and incubated for 1 h at 37 “C with 50 milliunits of enzyme. Samples were reprecipitated with methanol prior to loading on SDS gels.

RESULTS

Cell-associated and Secreted Forms of BAPP695 and pAPP751 in Transfected 293 Cells-Stable cell lines overex- pressing /?APP695 and BAPP75I were established. Cell ex- tracts and concentrated conditioned medium were analyzed by Western blot with antibodies directed against a portion of the extracellular BAPP domain (anti-BX5) or against a frag- ment extending from the fi-peptide through the C-terminal end of the molecule (anti-BX6). /3APP695 and pAPP751 were each represented by two major bands in these cells and by a single band in conditioned medium (Fig. IA). The apparent molecular masses of these bands estimated on the basis of Y!-labeled marker proteins (Amersham Corp.) are: for pAPP695, 98 and 115 kDa in cells and 102 kDa in medium; for bAPP751, 108 and 130 kDa in cells and 115 kDa in medium. Other marker proteins (Bio-Rad) led to approxi- mately 10 kDa higher estimates. While the cell-associated @APP forms were recognized by both anti-BX5 and anti-BX6, the secreted forms in medium were not detected by anti-BX6 (data not shown), indicating that they had lost the cyto- plasmic part of the molecule. This was confirmed by use of the previously described antiserum Cl directed against a peptide encompassing the C-terminal 20 residues of BAPP (Selkoe et al., 1988), which similarly recognized the cell- associated but not the secreted /3APP forms (data not shown). In addition to the high molecular mass bands of 98-130 kDa, anti-BX6 and anti-Cl also detected a band of strong immu- noreactivity at approximately 11.5 kDa (Fig. 1B) present in cell extracts which was not seen using anti-BX5. This 11.5- kDa band was seen in extracts of cells overexpressing either BAPP695 or /3APP751 but was not found in the medium. Its

MN anti-BX5 an&BX6 anti - BX6 FAW I 1 1 I

12 3 4 5 678

200-

- 200

116- - 42

92 -

FIG. 1. Western blot of ceIluIar and secreted forms of BAPP in 293 ceIIs and 293 ceIIs transfected with BAPP cDNAs. A, lanes 1 and 2, conditioned medium; lanes 3, 4, and 5, cellular extracts; lanes I and 3, 293 transfected with flAPP751; lanes 2 and 4,293 cells transfected with flAPP695; lune 5, untransfected 293 cells. 6.25% acrylamide gel. B, lanes 6-8 as 3-5 in A. 7.5-20% acrylamide gel. Note the presence of endogenous OAPP 751/770 in non-transfected cells and the immunoreactive band at 11.5 kDa (arrow) in both transfected and nontransfected cells.

by guest on April 23, 2018

http://ww

w.jbc.org/

Dow

nloaded from

4494 Alzheimer BAPP Metabolism

reactivity with anti-BX6 indicates that it is derived from the C terminus of /3APP. Three antisera raised against synthetic peptides containing amino acids l-38 and l-43 of the p- peptide and two antisera raised against HPLC-purified amy- loid plaques from Alzheimer brain recognized neither the 11.5 kDa fragment nor full length or secreted @APP in Western blot (data not shown), although these antisera react strongly with amyloid plaque protein in immunohistochemistry.

Time Course of @APP Synthesis-To investigate the time course of appearance of the various @APP immunoreactive bands in the 293 cell transfectants, cells were pulse-labeled for 5 min with [35S]methionine and chased with cold methio- nine-containing medium for various lengths of time, after which cell extracts and conditioned medium were harvested and immunoprecipitated with anti-BX6. When cells were harvested immediately after pulse labeling without a chase period, cell extracts contained a single /3APP-specific band corresponding to the lower molecular weight band of the two seen in Western blots (Fig. 2). After 40 min of chase the higher molecular weight form appears, and after 120 min nearly all of the radioactivity was in the higher molecular weight form. Experiments with longer chase periods (Fig. 3) demonstrated that by 4 h both the higher and lower molecular weight forms were virtually gone. Time courses did not differ significantly between bAPP695 and pAPP751. Identical re- sults were obtained when anti-BX5 was used for immunopre- cipitation (data not shown). In conditioned medium from pulse-chase experiments, a single band was detected by anti- BX5 after 40-60 min of chase. The intensity of this band reached a maximum after approximately 2 h and was mark- edly decreased after 4 h (Fig. 2). The upper band (Fig. 2, arrow) which was immunoprecipitated from conditioned me- dium of pAPP695 overexpressing cells at the 120-min time point represents the endogenous form of pAPP751/770 (Tan- aka et al., 1988; Weidemann et al., 1989), indicating that its rate of secretion is similar to that of @APP695. This was confirmed with cells transfected with a /?APP75l expression construct (data not shown). Immunoprecipitation was specific since the relevant bands could not be immunoprecipitated after preabsorption of the antibodies with the appropriate bacterial i%sion protein (Fig. 3, ckl.).

Posttranslational Modifications-The existence of two

tvw ceils, anti - BX6 cond. medium, anti-BX5 !v%fv

1 I I I

0 20 40 60 120 min 20 40 60 120 240 min

200- - 200

92 - ..-*J?&S-

FIG. 2. Time course of BAPP695 synthesis and secretion in transfected 293 cells. Cells were pulse-labeled with [%]methio- nine for 5 min; cells were lysed and conditioned medium collected after chase periods as indicated. /3APP was immunoprecipitated with anti-BX6 from cellular extracts and with anti-BX5 from conditioned medium. Note the presence of endogenous BAPP751/770 (arrow) in conditioned medium. 7.5-20% acrylamide gel. 20 min, increased in intensity over 2 h, and was still present

ctrl. lhr 2hrs 4hrs 6hrs Ehrs

6APP695 anti-BX6

6APP751 w- w

anti-BX6

- 200

- 92

- 69

- 46

- 200

- 92

- 69

- 46

FIG. 3. Time course of bAPP695 and BAPP751 synthesis in transfected 293 cells. Cells were pulse-labeled for 10 min and chased with cold medium for lengths of time as indicated before immunoprecipitation with anti-BX6. Control lane (cd) immunopre- cipitation of a sample identical t,o that in the l-h time point, but using anti-BX6 preabsorbed with bacterial fusion protein BX6. 7.5- 20% acrylamide gel. The band at 200 kDa in all lanes is nonspecific.

bands derived from each /YAPP cDNA expression construct suggested that the protein is subject to posttranslational modification. The early appearance of a lower molecular weight band followed by the appearance of a higher molecular weight band further suggested that these were immature and mature forms of full length BAPP, respectively. We investi- gated this directly by use of sugar-cleaving enzymes and biosynthetic labeling. Treatment of immunoprecipitated /3APP695 and bAPP751 from cell extracts with endoglycosi- dase F decreased the molecular mass of the immature (98 and 108 kDa) bands by approximately 2 kDa (Fig. 4A), indicating the presence of asparagine-linked carbohydrate. Endoglyco- sidase H treatment led to similar results (data not shown). The mature (115 and 130 kDa) bands, however, showed no detectable decrease in molecular mass upon endoglycosidase F treatment. They were markedly reduced in size, however, when treated with neuraminidase, as were the secreted forms of /?APP (Fig. 4B). This indicates the presence of a large number of sialic acid residues in the mature &APP molecules. Treatment with fucosidase, N-acetyl@-D-glucosaminidase, endo-a-N-acetylgalactosaminidase and combinations of these enzymes were without detectable effect (data not shown). Biosynthetic labeling with [3zP]phosphate and [3%]sulfate demonstrated that the /3APP incorporates both labels (Fig. 4, C and D). These labels were only seen in the mature forms of @APP. This shows that DAPP is a phosphoprotein and is also in accord with reports of the presence of tyrosine sulfate in the molecule (Schubert et al., 1989; Weidemann et al., 1989).

Detection of a Stable 11.5-kDa Intermediate in the Process- ing of /3APP-The 11.5-kDa fragment seen in Western blots of 293 cell extracts with anti-BX6 (Fig. 1B) was also detected in pulse-chase experiments. It appeared in cell extracts after

by guest on April 23, 2018

http://ww

w.jbc.org/

Dow

nloaded from

Alzheimer BAPP Metabolism

6APP695 +neuramin. w -Endo +En& - +neuramin.

751 695 751 695 ’ F F 751 695’ 751 695 Carl. 695 695 Gtd. w

- 200

zi5 S-met condiiioned medium cells 35 SD4 32~~ antLBX6 anti-BX5 anGBX6 arttLBX6

A B C D FIG. 4. Posttranslational modifications of @APP. A, immunoprecipitation of OAPP from fiAPP695 trans-

fected 293 cells after [35S]methionine labeling for 1 h, with or without endoglycosidase F treatment. Note molecular weight shift of the immature forms of both transfected @APP695 (b to b’) as well as endogenous /3APP751 (u to a’). 6.25% acrylamide gel. B, Western blots of 293 cells and conditioned medium transfected with bAPP695 and pAPP751 after treatment with neuraminidase (neururmn.). Neuraminidase digestion of cell extracts (B, right pane/), although not complete, reduced the intensity of the mature bands and generated material of a molecular weight intermediate between that of the immature and mature forms. The immature forms are unaffected. 6.25% acrylamide gel. C and D, immunoprecipitation of bAPP695 from transfected HS 683 cells after metabolic labeling with [35S]sulfate and [‘*P]phosphate. Only mature 115-kDa pAPP695 is labeled, although in this experiment the mature protein is resolved into two distinct bands. Control lanes (&FL), IgG fractions of preimmune serum. 7.5- 20% acrylamide gel.

in considerable amounts after 8 h, when full length BAPP had disappeared completely from the cell (Fig. 5). No significant differences in the rate of formation of this fragment were seen between pAPP695 and bAPP751 overexpressing cells (Fig. 5). This fragment was not detected by anti-BX5, and preabsorp- tion of anti-BX6 with BX6 fusion protein abolished immu- noprecipitation of the fragment, demonstrating the specificity of the anti-BX6 reaction. The molecular mass of approxi- mately 11.5 kDa, which corresponds to about 100 amino acids, and the fact that the fragment is recognized by antibodies against the C terminus of BAPP but not by antibodies against a /3APP fragment directly N-terminal of the &peptide region, suggest that the N terminus of this fragment lies in the vicinity of the N terminus of the @-peptide. One should expect that the location of this cleavage site relative to the p-peptide could be clarified by antibodies directed against the B-peptide. A total of six such antisera has been tested in immunoprecip- itation: two peptide antisera raised against amino acids l-38 of the b-peptide derived from two different laboratories, two antisera raised against amino acids l-43, and two antisera raised against HPLC-purified amyloid plaque protein from Alzheimer post-mortem brain. All of these antibodies recog- nize amyloid plaques in immunohistochemistry. However, as in the Western blot experiments described above none of them recognized full length /3APP, secreted /3APP, or the 11.5-kDa fragment even when protein extracts were boiled in 0.5% SDS before immunoprecipitation to uncover masked epitopes (data not shown).

DISCUSSION

The accumulation of the /3-peptide in amyloid plaques suggests that the normal metabolism of /3APP has been al- tered in Alzheimer’s disease brain. We have used cell lines stably transfected with /3APP cDNA expression constructs to analyze the normal metabolism of @APP and to look for steps

that may be intermediates in the amyloidogenic pathway. In pulse-chase experiments with human embryonic kidney 293 cells separately transfected with fiAPP695 and /3APP751 cDNA expression plasmids, each transfected cDNA led to two major forms of DAPP detectable inside cells: an immature, endoglycosidase F-sensitive species which was synthesized early, and a higher molecular weight, mature neuraminidase- sensitive form which appeared over a 2-h period. Analysis of culture supernatants revealed that for each /3APP form there was an anti-BX5 immunoreactive protein that was lower in molecular weight than the respective mature membrane-as- sociated form. This secreted material was first seen in the medium after 40 min of cold chase and peaked at approxi- mately 2 h. A similar sequence of events has been established recently for /3APP metabolism in HeLa cells (Weidemann et al., 1989). In our experiments, we have identified an additional 11.5-kDa band that appeared during the chase period in the cell pellets, with a time course similar to that seen in the medium for secreted material. These results suggest that for each of @APP695 and pAPP751, an immature N-glycosylated protein is rapidly synthesized, which is then processed to a higher molecular weight, mature, fully glycosylated form. This mature form is then cleaved to simultaneously yield a secreted extracellular domain and a long-lived, 11.5-kDa, membrane- bound fragment. This cleavage event probably occurs also in brain, since an approximately 11-kDa band has been seen in rat and human cerebral cortex with @APP C-terminal anti- bodies (Selkoe et al., 1988) and bands of 91 and 112 kDa have been seen in human cerebrospinal fluid with antibodies against the extracellular part of /3APP (Palmert et ul., 1989; Weidemann et ul., 1989). The cleavage of membrane-bound precursors to yield secreted products has previously been described for several proteins including epidermal growth factor (Doolittle et cd., 1984), transforming growth factor a (Bringman et cd., 1987), and rat liver sialyltransferase (Paul- son et ul., 1987).

by guest on April 23, 2018

http://ww

w.jbc.org/

Dow

nloaded from

4496 Alzheimer PAPP Metabolism

0APP751 BAPP695

clrl. lhr Zhrs 4hrs 6hrs 6hrs CM. lhr 2hrs 4hrs 6hrs 6hrs !#i

- 3.4

A B Clrl. ihr 2hrs 4hrs 6hrs 6hrs 0 20 40 60 120 mln

c D

FIG. 5. 11.5-kDa fragment of /3APP. Immunoprecipitation after a lo-min pulse label with [35S]methionine and indicated chase periods. A, 293 cells transfected with j3APP751; the antibody used was anti-BX6. Two nonspecific bands are seen in each lane, including the preabsorbed control lane (CM). The specific 11.5kDa band is the lowest of the three bands and is seen at all time points but absent in the control lane. I3,293 cells transfected with BAPP695; the antibody used was anti-BX6. As in A, the lowest band is the specific 11.5-kDa band. C, same as A but probed with anti-BX5. Note that anti-BX5 reacts only with the nonspecific bands also present in the control lane and does not recognize the 11.5-kDa band. LI, 293 cells trans- fected with flAPP695 and chased for shorter periods of time; 11.5- kDa fragment visible in small amounts after 20 min. CM, control lanes where samples identical to those of the l-h time point were immunoprecipitated with antibodies preabsorbed with corresponding bacterial fusion proteins BX5 and BX6.

Several considerations further support the conclusion that the fully glycosylated /3APP is the precursor to the cleaved 11.5kDa and secreted forms. First, although each @APP expression plasmid generated two cell-associated forms, each yielded only one secreted form. These secreted forms, which by immunochemical criteria have lost a C-terminal fragment, have an estimated molecular mass 10-15 kDa lower than the mature BAPP, in accord with the proposed precursor-product relationship. In contrast, the secreted forms are larger than their respective immature cell-associated @APP forms. Sec- ond, the secreted forms show a molecular weight shift upon neuraminidase treatment that is comparable to that seen upon enzyme treatment of the mature /lAPP (Fig. 5), whereas the immature form is unaffected by such treatment. And third, bands corresponding in mobility to those seen in the medium were never observed in immunoprecipitation of cell pellets. This suggests that the cleavage which generates the secreted form occurs either very late in the intracellular secretion pathway or on the cell surface. This is in accord with cleavage occurring after late posttranslational modifications have been made in the trans-Golgi compartment.

The metabolism of the long-lived 11.5kDa fragment may have special relevance to the mechanism of amyloidosis, since its size suggests that it may contain the fl-peptide. It must contain the transmembrane domain and thus at least part of the fl-peptide and the C terminus since it is an integral membrane fragment and reacts with C-terminal antibodies. Antibodies directed against the /?-peptide were not helpful in identifying exactly how much of the @-peptide is part of the

11.5-kDa fragment. A total of eight such antisera were tested in Western blot and in immunoprecipitation and were found not to react with full length @APP, secreted /3APP, or the 11.5-kDa fragment. This lack of reactivity may reflect speci- ficity for conformational epitopes only present in aggregated amyloid as all of these antisera recognize amyloid plaques in brain. Clarification of the exact location of the cleavage site will require purification and sequencing of the fragment.

We have shown that the 11.5-kDa fragment is quite stable in membranes. The pulse-chase experiments showed that it accumulated over a 2-h period and that substantial amounts remained after 8 h of chase (Fig. 5). Western blots of cell pellets, which reflect the steady state accumulation of protein in a cell, show substantial accumulation of this fragment (Fig. 1). Similarly, the 11.5-kDa fragment is readily detectable in postmortem brain and rat brain (Selkoe et al., 1988). The stability and accumulation of the 11.5-kDa fragment raises the possibility that it may be an intermediate in the process of amyloid deposition. Indeed, it has been shown that a similar sized fragment created by in uitro translation aggregates in the absence of membranes and that aggregates can be digested by addition of proteinase K to yield a 4-kDa monomer (Dyrks et aZ., 1988).

Biosynthetic labeling of pAPP695 and pAPP751 showed that both are sulfated and phosphorylated in HS 683 cells. These results are in accord with previous observations that /3APP undergoes a complex set of posttranslational modifi- cations. Sulfate has been demonstrated in /3APP secreted from PC12 cells and HeLa cells and shown to be covalently attached to tyrosine in PAPP (Schubert et al., 1989; Weide- mann et al, 1989). Regarding phosphorylation, previous work has shown that synthetic peptides derived from the @APP sequence can serve as substrates for protein kinase C and Ca*+/calmodulin-dependent protein kinase II (Gandy et aZ., 1988). Further work will be necessary to assess the relation- ship of these synthetic peptide results to our observations that DAPP is a phosphoprotein, since we do not know the identity of the kinase(s) which act on DAPP in cells. It is of interest to note, however, that a possible site for tyrosine phosphorylation similar to that seen in the epidermal growth factor receptor is present at residue 687 in the cytoplasmic domain of /3APP, but we do not know if this site is actually phosphorylated.

The susceptibility of @APP to endoglycosidase F, endogly- cosidase H, and neuraminidase treatments indicates that it contains covalently bound sugars. The immature cell-associ- ated form showed an apparent molecular mass decrease of approximately 2 kDa upon treatment with either endoglyco- sidase F or H, indicating the the presence of N-linked high mannose sugars. This confirms earlier work which showed that tunicamycin treatment caused a similar shift in mobility of /3APP (Weidemann et al., 1989) and is in accord with in uitro translation results (Dyrks et al., 1988). We could not observe any effect of endoglycosidase F on mature /3APP. This may be due to maturation of N-linked sugars to a complex form not cleaved by endoglycosidase F, or, alterna- tively, it is possible that a small molecular weight shift was not resolved in our gel system for this larger protein. This larger form of /3APP, however, was susceptible to neuramini- dase treatment, indicating the presence of multiple sialic acid residues. Although sialic acid can be present in both N-linked and O-linked sugar side chains, the endoglycosidase F resist- ance suggests that it is present in O-linked forms. In some instances as in the neuronal cell adhesion molecule N-CAM, attachment of large polysialic acid chains to N-linked glycans can cause a considerable shift in molecular weight (Rothbard

by guest on April 23, 2018

http://ww

w.jbc.org/

Dow

nloaded from

Alzheimer BAPP Metabolism 4497

et al., 1982). In N-CAM, however, these sugars can be removed completely by endoglycosidase F (Cunningham et al., 1983). The observation that tunicamycin does not block the shift in mobility from immature to mature forms (Weidemann et al., 1989) also argues that maturity involves addition of O-linked sugars. Further work will be necessary to assess the potential importance of posttranslational modifications in the control of @API’ turnover and in the processes that lead to b-peptide aggregation and amyloid deposition.

Acknoudedgments-We are thankful to Walter R6wekamp, Hei- delberg, for providing the expression vectors pExlOmer and pExl2mer, Hans-Ulrich Bernard, Singapore, for the expression vec- tors of the pORFex series, Dennis Selkoe for various antibodies against HPLC-purified and synthetic @-amyloid peptides, and Dale Schenk for helpful discussions.

REFERENCES

Bringman, T. S., Lindquist, P. B., and Derynck, R. (1987) Cell 48, 429-440

Cunningham, B. A., Hoffman, S., Rutishauser, U., Hemperley, J. J., Edelman, G. M. (1983) Proc. N&l. Acad. Sci. U. S. A. 80, 5762- 5766

de Sauvage, F., and Octave, J. N. (1989) Science !245,651-653 Doolittle, R. F., Feng, D. F., and Johnson, M. S. (1984) Nutwe 307,

558-560 Dyrks, T., Weidemann, A., Multhaup, G., Salbaum, J. M., Lemaire,

H.-G., Kang, J., Mtiller-Hill, B., Masters, C. L., and Beyreuther, K. (1988) EMBO J. 7,949-957

Gandy, S., Czernik, A. J., and Greengard, P. (1988) Proc. N&l. Acud. Sci. U. S. A. 85,621%6221

Glenner, G. G., and Wong, C. W. (1984a) &o&em. Biop/~ys. Res. commun. 120,885-890

Glenner, G. G., and Wong, C. W. (1984b) &o&em. Biophys. Res. cormnun. 122,1131-1135

Goldgaber, D., Lerman, M. I., MC Bride, 0. W., Saffiotti, U., and Gajdusek, D. C. (1987) Science 235,877-880

Kang, J., Lemaire, H.-G., Unterbeck, A., Salbaum, J. M., Masters, C. L., Grzeschik, K.-H., Multhaup, G., Beyreuther, K., and Mtiller- Hill, B. (1987) Nutwe 325, 733-736

Katzman, R. (1986) N. Engl. J. Med. 314,964-973 Kitaguchi, N., Takahashi, Y., Tokushima, Y., Shiojiri, S., and Ito, H.

(1988) Nuture 33 1, 530-532 Masters, C. L., Simms, G., Weinman, N. A., Multhaup, G., MC

Donald, L.A., and Beyreuther, K. (1985) Proc. Nutl. Acud. Sci. U. S. A. 82,4245-4249

Oltersdorf, T., Fritz, L. C., Schenk, D. B., Lieberburg, I., Johnson- Wood, K. L., Beatie, E. C., Ward,. P. J., blather, R.-W.,.Dovey, H. F.. and Sinha, S. (1989) Nuture 341,144-147

Paliert, M. Rl, PAdlisiy, M. B., gitker, D. S., Oltersdorf, T., Younkin, L. H., Selkoe, D.J., and Younkin, S. G. (1989) Proc. Nutl. Acud. Sci. U. S. A. 86,6338-6342

Paulson, J. C., Weinstein, J., Ujita, E. L., Riggs, K. J., and Lai, P.H. (1987) Biochm. Sot. Truns. 15,618-620

Ponte, P., Gonzalez-DeWhitt, P., Schilling, J., Miller, J., Hsu, D., Greenberg, B., Davis, K., Wallace, W., Lieberburg, I., Fuller, F., and Cordell, B. (1988) Nuture 331.525-527

Robakis, N. k., Rimakrishna, N., W&fe, G., and Wisniewski, H. M. (1987) Proc. Nutl. Acud. Sci. U. S. A. 84,4190-4194

Rothbard, J. B., Brackenbury, R., Cunningham, B. A., Edelman, G. M. (1982) J. Viol. &em. 257,11064-11069

Schubert, D., LaCorbiere, M., Saitoh, T., and Cole, G. (1989) Proc. N&l. Acud. Sci. U. S. A. 86, 2066-2069

Seedorf, K., Oltersdorf, T., Krhmmer, G., and Rtiwekamp, W. (1987) EMBO J. 6,139-144

Selkoe, D. J. (1989) Annu. Reu. Neurosci. 12, 463-469 Selkoe, D. J., Podlisny, M. B., Joachim, C!. L., Vickers, E. A., Lee, G.,

Fritz, L. C., and Oltersdorf, T. (1988) Proc. N&l. Acud. Sci. U. S. A. 8&734i-7345

Tanaka, S., Nakamura, S., Ueda, K., Kameyama, M., Shiojiri, S., Takahashi, Y., Kitaguchi, N., and Ito, H. (1988) Bioc!zern. Eiophys. Res. Commun. 157,472-479

Tanzi, R. E., Gusella, J. F., Watkins, P. C., Bruns, G. A. P., St. George-Hyslop, P. H., VanKeuren, M. L., Patterson, D., Pagan, S., Kurnit, D. M., and Neve, R. L. (1987) Science 235, 880-884

Tanzi, R. E., McClatchey, A. I., Lamperti, E. D., Villa-Komaroff, L., Gusella, J. F., and Neve, R. L. (1988) Nuture 331, 528-530

Van Nostrand, W. E., Wagner, S. L., Suzuki, M., Choi, B. H., Farrow, J. S., Geddes, J. W., Cotman, C. W., Cunningham, D. D. (1989) Nuture 341,546-549

Weidemann, A., Khnig, G., Bunke, D., Fischer, P., Salbaum, J. M., Masters, C. L., and Beyreuther, K. (1989) Cell 57, 115-126

by guest on April 23, 2018

http://ww

w.jbc.org/

Dow

nloaded from

T Oltersdorf, P J Ward, T Henriksson, E C Beattie, R Neve, I Lieberburg and L C Fritzin the biosynthetic/degradative pathway.

The Alzheimer amyloid precursor protein. Identification of a stable intermediate

1990, 265:4492-4497.J. Biol. Chem. 

  http://www.jbc.org/content/265/8/4492Access the most updated version of this article at

 Alerts:

  When a correction for this article is posted• 

When this article is cited• 

to choose from all of JBC's e-mail alertsClick here

  http://www.jbc.org/content/265/8/4492.full.html#ref-list-1

This article cites 0 references, 0 of which can be accessed free at

by guest on April 23, 2018

http://ww

w.jbc.org/

Dow

nloaded from