expression of the chondroitin sulfate proteoglycans of amyloid precursor (appican) and amyloid...

Journal of NeurochemistryLippincott-Raven Publishers, Philadelphia© 1995 International Society for Neurochemistry

Expression of the Chondroitin Sulfate Proteoglycans ofAmyloid Precursor (Appican) and Amyloid

Precursor-Like Protein 2

Menelas N. Pangalos, Junichi Shioi, and Nikolaos K. Robakis

Department of Psychiatry and Fishberg Research Center for Neurobiology,Mount Sinai School ofMedicine, New York, New York, U.S.A .

Abstract: The Alzheimer amyloid precursor (APP) proteinis a member of a family of glycoproteins that includes theamyloid precursor-like proteins (APLPs) . Previously, weshowed that in C6 glioma cell cultures, secreted APPnexin II occurs as the core protein of a chondroitin sulfateproteoglycan (CSPG) . Here, we report that among sevenuntransfected cell lines, expression of secreted APPCSPG was restricted to two cell lines of neural origin,namely, C6 glioma and Neuro-2a neuroblastoma (N2a)cells . Addition of dibutyryl cyclic AMP in N2a cultures, atreatment that induces the neuronal phenotype in thesecells, resulted in a significant reduction in the amount ofthe secreted APP CSPG, although secretion of APP wasonly marginally affected . Growth in the presence of serumincreased the size of the secreted APP CSPG, suggestingthat the number and/or length of the chondroitin sulfate(CS) chains attached to the core APP varies with growthconditions . Extensive mapping with epitope-specific anti-bodies suggested that a CS chain is attached within orproximal to the A/3 sequence of APP . In contrast to therestricted expression of the APP CSPG, expression ofsecreted APLP2 CSPGs was observed in all cell linesexamined . After chondroitinase treatment, two core pro-teins of -100 and 110 kDa were obtained that reactedwith an APLP2-specific antiserum, suggesting that non-transfected cell lines contain at least two endogenousAPLP2 CSPGs, probably derived by alternative splicingof the APLP2 KPI domain . The fraction of the APLP2proteins in the CSPG form was dependent on the particu-lar cell line examined . The proteoglycan nature of APPand APLP2 suggests that addition of the CS glycosami-noglycan chains is important for the implementation ofthe biological function of these proteins . However, thedifferential expression of these two proteoglycans sug-gests that their physiological roles and their possibleinvolvement in Alzheimer's disease may differ . KeyWords : Alzheimer's disease-Chondroitin sulfate pro-teoglycans-Alzheimer amyloid precursor-Amyloidprecursor-like proteins .J. Neurochem. 65, 762-769 (1995) .

Alzheimer amyloid precursors (APPs) are importantproteins in Alzheimer disease (AD), because they arethe precursors of A,6 peptide, which aggregates to formthe amyloid depositions characteristic of AD pathology

762

(Goldgaber et al ., 1987 ; Kang et al ., 1987 ; Robakiset al., 1987 ; Tanzi et al ., 1987), and because certainmutations of the APP gene appear sufficient for theinduction of the AD phenotype (Goate et al ., 1991 ;Murrell et al ., 1991 ; Mullan et al ., 1992) . At leastthree APP isoforms have been identified containing695, 751, or 770 amino acid residues (Kitaguchi etal ., 1988) . Full-length APPs are type I transmembraneglycoproteins with a large extracytoplasmic region anda small cytoplasmic domain . The A)3 sequence in-cludes the last 28 extracytoplasmic residues and 11-15 amino acids of the adjacent transmembrane domain.APPs are extensively modified posttranslationally, aprocess that may affect both the production of A)3 andthe biological function of these proteins . Nonamy-loidogenic secreted APP, containing almost all of theextracytoplasmic sequence but without the transmem-brane and cytoplasmic regions, is produced whenmembrane full-length APP is cleaved by the unidenti-fied enzyme "a-secretase" within the A,Q sequence .Cleavage outside the A,Q sequence may produce amy-loidogenic secreted or membrane-bound APP frag-ments (for review, see Robakis, 1994) . APPs are ex-pressed in almost all tissues examined, including brain .They have been suggested to have an array of biologi-cal functions including stimulation of cell growth (Sai-toh et al ., 1989), cell adhesion (Schubert et al ., 1989a ;Breen et al ., 1991) , stimulation of neurite outgrowth(Robakis et al ., 1990 ; Jin et al ., 1994), and neuropro-tective activities (Mattson et al ., 1993) .

Received November 16, 1994 ; revised manuscript received Janu-ary 20, 1995 ; accepted January 27, 1995 .

Address correspondence and reprint requests to Dr. N. K . Robakisat Mount Sinai School of Medicine, Department of Psychiatry andFishberg Research Center for Neurobiology, One Gustave L . LevyPlace, New York, NY 10029, U.S.A .

Abbreviations used: aa, amino acid ; AD, Alzheimer disease ; APP,Alzheimer amyloid precursor ; CAMP, cyclic AMP ; CHO, Chinesehamster ovary ; CS, chondroitin sulfate ; CSPG, chondroitin sulfateproteoglycan; DMEM, Dulbecco's modified Eagle's medium ; FBS,fetal bovine serum; GAG, glycosaminoglycan ; KPI, Kunitz proteaseinhibitor ; mAb, monoclonal antibody ; N2a, mouse Neuro-2a neuro-blastoma ; SDS, sodium dodecyl sulfate .

Protein purification and sequencing studies showedthat a novel chondroitin sulfate proteoglycan (CSPG),secreted by C6 cells and migrating on sodium dodecylsulfate (SDS)-polyacrylamide gel electrophoresis be-tween 150 and 250 kDa, had the Kunitz-type serineprotease inhibitor (KPI) -containing, 120-kDa secretedAPP as its core protein (Shioi et al ., 1992, 1993) .CSPGs are a class of molecules consisting of a coreprotein to which one or more chondroitin sulfate (CS)glycosaminoglycan (GAG) chains are covalentlyattached . The CS GAG chains are bound via a xylosesugar to a serine residue of the core protein (Loh-mander et al ., 1989) . Due to the large size of the GAGchains, proteoglycans can be considerably larger thantheir core proteins . These macromolecules are postu-lated to be involved in a number of key cellular eventsincluding cell adhesion, cell-cell communication,modulation of growth factor activities, and binding ofapolipoproteins (Ruoslahti, 1988; Jackson et al.,1991) . In the brain CSPGs have been suggested tohave neuroprotective properties and to modulate axo-nal growth and neural patterning (Snow et al ., 1990 ;Brittis et al ., 1992 ; Margolis and Margolis, 1993; Oka-mato et al ., 1993, 1994) . Proteoglycans, including he-paran, dermatan, and CSs, have been found in andaround senile plaques and neurofibrillary tangles andhave been suggested to play a role in the pathogenesisof AD (Snow et al ., 1992, 1994a ; Su et al ., 1992 ;DeWitt et al ., 1993) .

Recently, it was reported that APLP2, another mem-ber of the APP family of proteins, when transfectedinto Chinese hamster ovary (CHO) cells or COS 1 fi-broblasts, was expressed as the core protein of a CSPG(Thinakaran and Sisodia, 1994) . APLPs display highhomology with the APP, but they do not contain theA,3 sequence (Sprecher et al ., 1993 ; Wasco et al .,1993) and their role in AD remains unclear .

In the present study, we investigated a number ofuntransfected cell lines to determine the endogenousexpression of APP and APLP2 CSPGs using specificantisera . We found that expression of APP CSPG wasrestricted to cell lines of neural origin, whereas APLP2CSPG was expressed ubiquitously by all cell lines in-vestigated . Evidence was also obtained suggesting thatgrowth conditions may affect the size or number ofGAG chains attached to APP and that secretion of APPCSPG may be regulated differently from that of APP .Furthermore, in contrast to recent reports of immuno-cross-reactivity between the two core proteins of theseCSPGs (Thinakaran and Sisodia, 1994; Slunt et al .,1994), several APP antibodies used in this studyshowed no cross-reactivity with the APLP2 CSPG andwere used to distinguish between the expression ofthese two molecules .

MATERIALS AND METHODSCell cultures and treatmentsHuman HTB14 glioblastoma (HTB14), rat C6 glioma

(C6), mouse Neuro-2a neuroblastoma (N2a), human SY5Y

EXPRESSION OF APP AND APLP2 PROTEOGLYCANS 763

neuroblastoma (SY5Y), human 293 embryonic kidney fi-broblasts (293), monkey COS kidney fibroblasts (COS),and CHO cells were obtained from the American Type Cul-ture Collection bank. Cultures were initially grown in Dul-becco's modified Eagle's medium (DMEM; GIBCO) sup-plemented with 10% (vol/vol) fetal bovine serum (FBS;JRH Biosciences) . Upon reaching confluency, cells werewashed with DMEM and then cultured for 3-4 days inDMEM media containing 1 mg/ml transferrin, 5 hg/ml insu-lin, 30 nM sodium selenite, 20 nM progesterone, and 100 MMputrescine . At the end of the incubation period, conditionedmedia were collected and processed as indicated in eachexperiment. In certain experiments, N2a cultures at 60-70%confluency were washed with DMEM and then placed for3-4 days in either (1) DMEM supplemented with 10% FBS,(2) DMEM with 10% FBS containing 1 mMdibutyryl cyclicAMP (cAMP) (Sigma), or (3) DMEM plus nonserum sup-plements (see above) . Cultures were maintained at 37°C,in an atmosphere of 95% air, 5% COZ , and 100% relativehumidity. Detergent lysates of Sf9 cells expressing full-length mouse APLP2 were kindly provided by Dr. S . Sisodia .

Chondroitinase digestionsAliquots of media were subjected to digestion with 0.01

units of protease-free, chondroitinase ABC (EC 4.2 .2 .4,Boehringer Mannheim), an enzyme that degrades both CSand dermatan sulfate chains, in 200 p1 of 40 mM Tris-HCl(pH 8.0) and 20 mM sodium acetate (pH 8 .0) at 37°C for8 h . In some experiments, chondroitinase AC (EC 4.2 .2 .5,Sigma) was used, which specifically degrades CS chains .After digestion, samples were frozen, lyophilized to dryness,and then made 100 1A with 1 x Laemmli sample buffer .

Purification of APP CSPG from APPPurified APP CSPG free of APP was prepared as de-

scribed by Shioi et al . (1992) . In brief, conditioned mediumfrom cells grown in DMEM plus supplements was made 10mM in EDTA and then applied to a dextran sulfate-Sepha-rose column and the bound material eluted with a 0-1 .0 MNaCl gradient . Dialyzed fractions containing both APP andAPP CSPG were passed through an anion-exchange fastperformance liquid chromatography (FPLC) MonoQ HR5/5 column (Pharmacia) . APP was separated from APP CSPGusing a nonlinear salt gradient ranging from 0 to 1 M NaCl[0-0.25 M NaCl (2 ml), 0.25-0.65 M NaCl (18 ml), 0.65-1.0 M NaCl (10 ml), 1.0 M NaCl (l 1 ml) ] . Salt-containingfractions were dialyzed twice against 2 L of 20 mM Tris-HCl (pH 7.5) and 1 mM EDTA at 4°C before further use .Separation of APP CSPG from APP was determined byimmunoblotting .

SDS electrophoresis and immunoblotsAll samples were made to 1 x Laemmli sample buffer and

placed in boiling water for 5 min before electrophoresis on6 or 8% polyacrylamide minigels (Bio-Rad) in the presenceof 1% SDS . Proteins were electroblotted to polyvinylidenedifluoride membrane (Immobilon, Millipore) and immuno-detected with anti-APP or anti-APLP2 antibodies as de-scribed (Shioi et al ., 1992) . Cell protein concentration wasdetermined by BCA assay (Pierce) on cell homogenatesaccording to the manufacturer's instructions . Aliquots ofme-dia were run per gel lane to reflect equivalent amounts ofcell protein content . Quantification and densitometric mea-surements were performed with a computerized MagiscanJoyce Lobel image analysis system .

J. Neurochem., Vol . 65, No . 2, 1995

764

FIG. 1 . Antisera specificity to APP and APLP2. APP partiallypurified from the conditioned medium of C6 glioma cell culturesusing dextran sulfate column chromatography (lanes 1, 3, 5, and7) or extracts from Sf9 cells either transfected and expressingfull-length mouse APLP2 (lanes 2, 4, 6 and 8) or nontransfected(lane 9) were loaded onto a 6% denaturing polyacrylamide gel .After electrophoresis and transfer, membranes were probed withanti-APP or anti-APLP2 antisera. Lanes 1 and 2, GID; lanes 3and 4, R7 ; lanes 5 and 6, R47; lanes 7, 8, and 9, D21. Thickarrow marks position of APP and thin arrow marks position ofAPLP2. The lower molecular mass APP proteins, detected byGID, R7, and D21 antibodies, are likely degradation products ofAPP orAPLP2. Numbers at left represent mobilities of molecularmass markers in kilodaltons .

AntibodiesFor APP detection, the following antisera and monoclonal

antibodies (mAbs) were used [APP amino acid (aa) posi-tions are based on APP751 1 : rnAb 22C11 specific for as 46-61 (dilution 1 :200, Boehringer Mannheim), GID specific foras 179-185 (1:2,000 ; Schubert et al ., 1989b), R7 specificfor as 296-315, recognizing the KPI insert of APP (1 :4,000 ;Refolo et al ., 1989), R5 specific for as 476-496 (1 :1,000 ;Shioi et al ., 1992), rnAb 1G5 specific for as 500-556 (1/ug/ml, a gift from Athena Neurosciences), mAb Alz 90specific for as 567-664 (1 :200, Boehringer Mannheim), R3specific for as 628-652 (1:500 ; Refolo et al ., 1989), R47specific for as 652-667 (1 :1,000; Shioi et al ., 1992), andR1 specific for as 729-751 (1:1,000 ; Shioi et al ., 1993) . ForAPLP2 detection, D2I antisera was used specific for as 567-581 (1 :6,000 ; a kind gift from Dr . S . Sisodia; see Thinakaranand Sisodia, 1994) .

RESULTS

Specificity of APP and APLP antiseraIt has been suggested, recently, that APP antibodies

may cross-react with APLP2 (Slunt et al ., 1994 ; Thina-karan and Sisodia, 1994) and that this may confoundidentification of the precise nature of the APP familyof proteoglycans . To examine the specificity of theantibodies used in this study, APP partially purifiedfrom the conditioned medium of C6 cell cultures, orAPLP2 from cell extracts of Sf9 cells transfected withAPLP2, were used . Antisera GID, R7, and R47, di-rected against different regions of APP (see Materialsand Methods), did not cross-react with APLP2 (Fig .1, lanes 1-6) . D2I antisera, directed against APLP2,was specific for APLP2 with no immunostaining ofAPP proteins (Fig . 1, lanes 7-9) . APP mAb Alz 90also failed to react with APLP2, whereas weak immu-noreactivity with APLP2 was observed with anti-APPantisera R5 (data not shown) .

J. Neurochem., Vol . 65, No. 2, 1995

M. N. PANGALOS ET AL.

Expression of CSPG APP and CSPG APLP2 indifferent cell lines

Conditioned media from seven different cell lineswere analyzed by immunoblotting using specific anti-sera GID and D21 . All cell lines studied secreted APPat 120 kDa to differing degrees (Fig . 2A, Table 1) .Under our conditions, HTB 14 and 293 cell culturessecreted very low levels of APP, detectable only whenhigher quantities of protein were loaded on the gel(data not shown) . Both C6 and N2a cells secretedan anti-APP-immunoreactive protein that resulted ina prominent diffuse band at 150-250 kDa, characteris-tic of the APP CSPG (Shioi et al ., 1992, 1993) . Thisband was eliminated by either chondroitinase ABC orAC treatment, with a concomitant increase in theamount of the 120-kDa core protein (Fig . 2A), indicat-ing that these proteoglycans contain APP as their coreprotein . Densitometric quantitation of the relativeamounts of the core protein before and after chondroi-tinase treatment indicated that in both cell cultures,>50% of the total secreted APP was in the CSPG form(Table 1) . Under these conditions, no CSPG APP wasdetected in CHO, COS, or SY5Y cultures (Fig . 2A),even when the x-ray film was exposed for longer timesto the blots so that the APP signal was comparablewith the APP signal from C6 and N2a cultures (datanot shown) . However, in 293 and HTB 14 cultures,which secrete very low levels of APP, we cannot ex-

FIG. 2. Secretion of APP CSPG and APLP2 CSPG by differentcell lines . Cultures from seven different cell lines were grown inDMEM plus defined supplements, and conditioned media werecollected after 3 days of confluent culture . Aliquots of mediawere incubated at 37°C without (-) or with (+) chondroitinaseABC before separation on an SDS-polyacrylamide gel and west-ern blotting . Volumes loaded were adjusted to account for differ-ences in total cellular protein present in each culture type, asdetermined by BCA assay of cell homogenates . Blots wereprobed with APP-specific GID antiserum (A) or APLP2-specificantiserum D21 (B) . To visualize expression of the CSPGs, whichare less reactive than the unmodified APP core proteins (Shioi etal ., 1992), x-ray films were overexposed and may not accuratelyreflect the relative expression of each molecule (see Table 1 forquantitation of CSPG expression) . HTB14, human glioblastoma,astrocytoma ; C6, rat glioblastoma ; SY5Y, human neuroblas-toma; N2a, mouse neuroblastoma; COS1, monkey kidney fibro-blast; 293, human embryonic kidney fibroblast ; and CHO. Molec-ular mass markers are in kilodaltons .

TABLE 1 . Secreted APP andAPLP2 CSPG expression in various cell lines

Samples were prepared as described in the legend to Fig. 2 and blots were probed with APP-specific GID or APLP2-specific D2I antisera .Densitometry was performed on autoradiograms in the linear range ofthe film from two independent experiments . Number of + signs indicatesrelative amounts of secreted APP or APP CSPG . (-), No detectable signal . Values given are average percentages of APP or APLP2 foundin the form of proteoglycan . Total APP(T) or APLP(T) is derived from the total amount of the respective core protein detected afterchondroitinase digestion. The amount of APP or APLP2 in the CSPG form is calculated by subtracting from APP(T) or APLP2(T) therespective protein signal obtained before chondroitinase treatment. The percentage of each APLP2 protein in the proteoglycan form wasdetermined separately for the 100- or 110-kDa core proteins .

clude the possibility that undetectable levels of APPCSPG were secreted.

All cell lines except 293 secreted significantamounts of two APLP2 proteins detected with D21antisera and migrating at -100 and 110 kDa . Themobility of these proteins was distinct from the mobil-ity of the secreted APP . In addition, D2I staining de-tected in the conditioned media of all cell lines asmeary signal between 120 and 200 kDa, characteristicof the heterogeneous migration of proteoglycans (Fig .2B) . This signal was strongest in HTB 14, COS, andCHO cells and suggests that they produce the highestamounts of this proteoglycan . Treatment of condi-tioned media from all cell cultures with chondroitinaseeliminated the heterogeneous smeary staining with apronounced concomitant increase in the amounts ofthe 100- and 110-kDa APLP2 proteins (Fig . 213) .These data are in agreement with a recent report thatan APLP2-transfected cell line secreted a CSPG con-taining the transfected protein in its core (Thinakaranand Sisodia, 1994) . However, the present results sug-gest that nontransfected cell lines secrete at least twoendogenous CSPGs containing either a 100- or 110-kDa APLP2 as their core protein . Densitometric analy-sis before and after chondroitinase treatment revealedthat the proportion to which each core protein wasfound in the proteoglycan form was dependent on theparticular cell line examined (Table 1) .

Determination of APP region containing the CSchainTo further characterize the possible sites of CS chain

attachment to the APP core protein, the APP CSPGwas separated from unmodified APP by applicationof conditioned medium to Sepharose-dextran sulfatefollowed by ion-exchange chromatography . This en-abled us to examine the immunoreactivity of the CSPGAPP and its resulting core protein after digestion byABC, without interference from any endogenously se-creted, unmodified APP . To determine whether puri-


fied APP CSPG also contained APLP2 CSPG, samplespurified from C6 cultures, which synthesize primarilythe KPI-containing APP (Shioi et al ., 1992), weretreated with chondroitinase . It can be seen in Fig . 3that this treatment of the purified sample yielded coreproteins reactive only with anti-APP antisera, whereasD2I failed to react with the core proteins, showing thatthis preparation contains no APLP2 proteoglycan(Fig . 3) .The purified CSPG APP preparation was mapped

extensively with epitope-specific antibodies . Four APPantibodies, R3, R47, 1G5, and Alz 90, specific to epi-topes proximal to or within the A,0 region of APP, werenot able to detect the heterogeneous smeary staining ofthe APP CSPG; whereas four antibodies, 22C11, GID,R7, and R5, specific to APP regions distant from theA,3 sequence, detected the APP CSPG smear clearly .(Figure 3 shows representative reactivities using anti-sera GID, R7, R47, and R3.) After chondroitinase di-gestion of the APP CSPG, all of the above antibodiesdetected the resultant APP core protein (Fig . 3) . Thesedata suggest that the GAG attachment site on the APPcore protein is near to or within the A,3 region, andbefore chondroitinase treatment the epitopes recog-

FIG. 3. Immunoblotting of enriched APP CSPG preparations .Medium from confluent C6 cultures was collected and APPCSPG was purified (see Materials and Methods) . Aliquots ofAPPCSPG were incubated at 37°C without (-) or with (+) chon-droitinase ABC before separation on an SDS-polyacrylamide geland western blotting . Blots were probed with APP-specific anti-sera GID, R7, R47, or R3 or with APLP2-specific antiserum D21.Molecular mass markers are in kilodaltons .

J. Neurochem., Vol. 65, No . 2, 1995

Cell type Species APPAPPCSPG

APP CSPG/APP(T)

APLP2 CSPG/APLP2(T) (110 kDa)

APLP2 CSPG/APLP2(T) (100 kDa)

HTB 14 ; glioblastoma, astrocytoma Human + - - 58 62C6 ; glial tumor cell Rat +++ +++ 54 34 62N2a ; neuroblastoma Mouse +++ +++ 74 82 76SY5Y ; neuroblastoma Human ++ - - 86 82293 ; kidney fibroblast Human + - - 71 58COS ; kidney fibroblast Monkey ++ - - 53 87CHO; ovary cell Hamster +++ - - 90 39

766

nized by the former group of antibodies were maskedby the highly charged GAG chains .

Regulation of CSPG APPTo investigate whether different culture conditions

could affect APP CSPG production, N2a cells weregrown in defined DMEM in the presence of either 10%FBS or supplements . Secreted APP CSPG, producedby cells grown in the presence of serum, had a highermolecular weight than APP CSPG produced by cellsgrown in chemically defined medium (Fig . 4) . Similarobservations were made with APP CSPG secreted byC6 glioma cells (data not shown) . Because no changein the size of the APP core protein was observed, theseresults suggest that serum factors affect the averagelength of the CS chain or chains and/or the numberof chains attached to the core protein .N2a cells were grown in DMEM media with 10%

FBS in the presence or absence of 1 mM dibutyrylcAMP, a treatment known to result in neuronal differ-entiation (Prasad, 1975) . In subconfluent cultures, weobserved that dibutyryl cAMP caused the characteristicappearance of multiple neurite projections from thecell body, confirming a change in the phenotype of thecells (data not shown) . Conditioned medium from N2acultures treated with 1 mM dibutyryl CAMP contained72 - 9% less APP CSPG relative to untreated cultures(Fig . 5), whereas APP secretion remained relativelyunaffected (12 - 12% reduction ; values are meansfrom five independent experiments ± SD) . The differ-ence between the secretion of APP CSPG and APPobserved under these conditions suggests that the ex-pression of these two molecules is regulated differentlyby the cell .

APP CSPGs (termed appicans, Shioi et al ., 1995)were first detected in C6 glioma cell cultures (Shioiet al ., 1992 ; 1993) . These proteoglycans are also foundin human and rat brain tissue as well as in rat brainprimary cell cultures, where they are expressed primar-ily by astrocytes (Shioi et al ., 1995) . The results pre-sented here show that secretion of appican depends onboth cell type and growth conditions . Among the celllines examined, C6 glioma and N2a neuroblastomacells produced high levels of the secreted soluble appi-can . In these cell lines >50% ofthe total secreted APPwas in the proteoglycan form . No secreted appican

J. Neurochem., Vol. 65, No. 2, 1995

FIG. 4. Effect of FBS on APP CSPG synthesis inN2a cells. Subconfluent N2a cultures were grownin DMEM in the presence (+) or absence (-) of10% FBSfor 3-4 days and analyzed as describedin Fig. 2 legend . Blots were probed with APP-specific antiserum R7 . Representative resultsfrom five experiments are shown.

DISCUSSION


was detected in any of the other cell lines examined.Treatment of N2a cells with the membrane-permeablecAMP analogue dibutyryl cAMP resulted in a dramaticreduction in the amount of secreted appican . DibutyrylcAMP promotes the neuronal phenotype in these cells,inducing the elaboration of neuronal processes (Pra-sad, 1975), suggesting that production of appican maybe reduced upon neuronal differentiation . This sugges-tion is in agreement with recent results showing thatin primary cultures, this molecule is produced by gliabut not neuronal cells (Shioi et al ., 1995) . Althoughtreatment of N2a cells with dibutyryl cAMP resultedin a pronounced reduction in the secretion of appican,APP secretion was only marginally affected, sug-gesting that these two molecules may be regulated bydifferent mechanisms .N2a cells grown in defined media secreted appican

of a lower molecular weight than that secreted by cellsgrown in serum, although the mobility of the core pro-teins remained constant. This observation suggests thatunidentified serum factors may affect either the sizeor the number of the GAG chains attached to the coreprotein, or both, and is consistent with reports that thenumber or size of the CS chains attached to certaincore proteins may be regulated by growth factors (Bas-sols and Massague, 1988) . The specificity in the pro-duction of the appican is also reflected in primary neu-ronal cultures, where this molecule is produced only byastrocytes, although APP is also expressed by neurons,microglia, and oligodendrocytes (Shioi et al ., 1995) .

Recently, it was reported that CHO or COS celllines transfected with APLP2 cDNA produced a CSPGthat contained transfected APLP2 as a core protein(Thinakaran and Sisodia, 1994) . Because the APLPproteins may be involved in AD (Wasco et al ., 1993 ;Thinakaran and Sisodia, 1994), we examined whetherAPLP2 proteins occur as CSPGs in nontransfected celllines . Our results indicate that in contrast to the restric-tive expression of the APP CSPG, all cell lines studiedsecreted at least two endogenous APLP2 CSPGs withcore proteins of 100 and 1 10 kDa . The fraction of eachAPLP2 protein in the proteoglycan form depended onthe particular cell line (see Table 1) . However, incontrast to recent suggestions of antibody cross-reac-tivity (Thinakaran and Sisodia, 1994), our resultsshow that the two proteoglycans can be clearly differ-entiated as most of the anti-APP antibodies probed didnot cross-react with the APLP2 protein . For example,

FIG. 5. Effect of dibutyryl CAMP on APPCSPG production in N2a cells. Subcon-fluent N2a cultures were grown in the ab-sence (-) or presence (+) of dibutyrylCAMP for 3-4 days and analyzed as de-scribed in Fig. 2 legend . Blots were probedwith APP-specific R7 antiserum. Repre-sentative duplicate resultsfrom five exper-iments are shown. Similar results were ob-tained using the monoclonal antibody22C11 .

antisera R3 and R47, which detected the core proteinof the APP CSPG, are directed against APP sequencesabsent from APLP proteins . Furthermore, antisera GIDand R7, which detected the APP CSPG, failed to detectthe APLP2 CSPG; whereas D2I, which recognizes theAPLP2 CSPG, did not detect the APP CSPG. Theseresults are in accord with protein purification and se-quencing studies used to establish the proteoglycannature of the APP (Shioi et al ., 1992) .

It appears unlikely that the 100-kDa protein is adegradation product of the 110-kDa protein, becauseit consistently appears in all cell lines examined, withno other apparent degradation products present in themedia. It is possible that these two core proteins arethe result of different glycosylation but appears moreconsistent with the recent finding that APLP2 existsas two distinct isoforms that derive from alternativesplicing of the KPI insert of the APLP2 (Wasco et al .,1993 ; Sandbrink et al ., 1994) . The differential regula-tion of the expression of the APP and APLP2 CSPGssuggests that these molecules may have different bio-logical functions . Furthermore, the presence of theAPP CSPG in a specific set of cell lines, comparedwith APLP2 CSPG, which appears to be expressedmore ubiquitously, suggests a cell-specific functionalrole for the APP CSPG.

Most proteoglycans characterized thus far have theirGAG chains attached to a serine residue that is part of aconsensus Ser-Gly dipeptide motif preceded by acidicresidues (Sunblad et al ., 1988 ; Kjellen and Lindahl,1991) . There are three Ser-Gly acceptor sequenceswithin APP-751 at positions 57, 637, and 660, but onlythe 57 and 660 serines are preceded by acidic residues .The latter serine is located within the A,6 sequence ofAPP. Extensive mapping with epitope-specific anti-bodies performed in this study on purified APP CSPG

EXPRESSION OF APP AND APLP2 PROTEOGLYCANS

767

FIG. 6. Antibody immunoreactivity to the APP CSPG before and after chondroitinase digestion in western blots. Schematic diagramshowing the epitope locations of the various APP antibodies used in this study. (+), Antibodies immunoreactive with the APP CSPGor APP core protein after chondroitinase digestion ; (-), antibodies unable to detect the intact APP CSPG . Arrows indicate positionsof the three serine residues that are candidates for GAG chain attachment . However, the consensus sequence at position 637 is notpreceded by acidic residues, unless exon 15 is spliced out as in L-APP isoforms (see text) . All antibodies except R1 were probedagainst secreted APP CSPG . Reactivities of R1 antibodies against full-length APP CSPG or its chondroitinase digest were taken froma previous study (Shioi et al ., 1993).

showed that antisera recognizing sequences distant tothe A,6 sequence of APP also recognized the APPCSPG. In contrast, antisera specific for areas near orwithin the A,(3 sequence were not able to detect theheterogeneous smear characteristic of the APP CSPG.These observations suggest that the large charge andsize of the CS chains mask the APP region recognizedby these antisera . Elimination of antibody reactivitydue to epitope masking by GAG chains is often ob-served in proteoglycans (Gowda et al ., 1989) . Afterchondroitinase digestion, the latter group of antiseradetected the resulting core protein (see Fig . 6), indicat-ing that digestion of the GAG chains removes thismasking, enabling detection of the protein by the anti-sera . The best consensus sequence for CS attachmentin the APP region masked by the GAG chains containsSer bb° , which corresponds to A,6 peptide residue 8,suggesting that a chain may be attached within the A(3sequence . Alternatively, splicing out of exon 15 ofAPP� o would create the L-APP mRNA (Konig et al .,1992), the translation product of which predicts a per-fect concensus sequence for the attachment of the CSchain at Ser b ' 9 of the resulting APP733 (Pangalos et al .,1995) . This serine is only 16 residues upstream of theN-terminus of the AO sequence of APP . It is attractiveto speculate that a GAG chain attached within thisregion affects the proteolytic processing of APP thatleads to the production of A,Q .

In the brain, CSPGs have been suggested to be in-volved in neuronal patterning (Snow et al ., 1990 ;Brittis et al ., 1992) and to participate in the healingprocess after brain injury (Okamato et al ., 1993,1994) . The involvement of proteoglycans in AD hasbeen suggested in a number of reports showing thepresence of the heparan sulfate proteoglycan perlecan(Su et al ., 1992 ; Snow et al ., 1994a,b), the dermatan


768

sulfate proteoglycan decorin (Snow et al ., 1992), and,most recently, the presence of CSPGs (DeWitt et al .,1993) in both senile plaques and neurofibrillary tan-gles . Furthermore, proteoglycans have been reportedto bind both APP and A0 (Narindrasorasak et al .,1991 ; Fraser et al ., 1992 ; Brunden et al ., 1993; Bueeet al ., 1993a,b) and to induce amyloid deposition inrodent brains (Snow et al ., 1994a) . That both APPand APLP2 CSPGs have been demonstrated to existas CSPGs has the following several important implica-tions for AD: (1) In AD there is a considerable neuriticregeneration activity (Geddes et al ., 1986) . BecauseCSPGs have been reported to modulate neurite out-growth, the possibility is raised that APP and APLPCSPGs may be involved in the neurite regenerationactivity of the AD brain. (2) The attachment of theGAG chains may change the proteolytic processing ofthese proteins . In the case of the APP CSPG this mayhave important implications for the production of amy-loidogenic APP fragments. (3) Because proteoglycanshave been shown to bind A0 (Fraser et al ., 1992 ;Brunden et al ., 1993 ; Buee et al ., 1993b), the CSPGform of these proteins may interact with A(3 in waysthat may result in either inhibition or stimulation ofA,6 aggregation. The detection of the APP and APLPproteoglycans suggests that these proteins have addi-tional biological functions deriving from their proteo-glycan nature . It is tempting to speculate that some ofthese functions may be involved in AD.

Acknowledgment: We thank Drs. James Ripellino andSpiros Efthimiopoulos for useful suggestions and Dr. San-gram Sisodia, Johns Hopkins University, for the kind gift ofD21 antisera and Sf9 cell extracts transfected with APLP2.This study was supported by NIH grants AG08200,AG05138, and by a grant from the A. P. Slaner family .

Bassols A. and Massague J. (1988) Transforming growth factor #regulates the expression of extracellular matrix chondroitin/dermatan sulfate proteoglycans. J. Biol. Chem . 263, 3039-3045.

Breen K. C., Bruce M., and Anderton B. H. (1991) 6 Amyloidprecursor protein mediates neuronal cell-cell and cell-surfaceadhesion. J. Neurosci . Res. 28, 90-100 .

Brittis P. A., Canning D. R., and Silver J. (1992) Chondroitin sulfateas a regulator of neuronal patterning in the retina . Nature 255,733-736.

Brunden K. R., Richter-Cook N. J., Chaturvedi N., and FredericksonR. C. A. (1993) pH-dependent binding of synthetic 0-amyloidpeptides to glycosaminoglycans, J. Neurochem. 61, 2147-2154.

Buee L., Ding W., Anderson J. P., Narindrasorasak S., KisilevskyR., Boyle N. J., Robakis N. K., Delacourte A., Greenberg B.,and Fillit H. M. (1993a) Binding of vascular heparan sulfateproteoglycan to Alzheimer's amyloid precursor protein is medi-ated in part by the N-terminal region of A4 peptide. Brain Res.627,199-204 .

Buee L., Ding W., Delacourte A., and Fillit H. (1993b) Binding ofsecreted human neuroblastoma proteoglycans to the Alzhei-mer's amyloid A4 peptide . Brain Res. 601, 154-163.

DeWitt D. A., Silver J., Canning D. R., and Perry G. (1993) Chon-droitin sulfate proteoglycans are associated with the lesions ofAlzheimer's disease . Exp. Neural. 121, 149-152.

Fraser P. E., Nguyen J . T., Chin D. T., and Kirschner D. A. (1992)


REFERENCES


Effects of sulfate ions on Alzheimer O/A4 peptide assemblies :implications for amyloid fibril-proteoglycan interactions . J.Neurochem. 59, 1531-1540.

Geddes J. W., Anderson K. J., and Cotman C. W. (1986) Senileplaques as aberrant sprout-stimulating structures . Exp. Neurol.94, 767-776.

Goate A., Chartier-Harlin M., Mullan M., Brown J., Crawford F.,Fidani L., Giuffra L., Haynes A., Irving N., James L., Mant R.,Newton P., Rooke K., Roques P., Talbot C., Pericak-Vance M�Roses A., Williamson R., Rossor M., Owen M., and HardyJ. (1991) Segregation of a missense mutation in the amyloidprecursor protein gene with familial Alzheimer's disease . Na-ture 349, 704-706.

Goldgaber D., Lerman M., McBride W., Saffioti U., and GajdusekD. (1987) Characterisation and chromosomal localization of aDNA encoding brain amyloid of Alzheimer's disease . Science235,877-880.

GowdaD. C., Margolis R. U., and Margolis R. K. (1989) Presenceof the HNK-1 epitope on poly (N-acetyllactosaminyl) oligosac-charides and identification of multiple core proteins in the chondroitin sulfate proteoglycans of brain. Biochemistry 28, 4468-4474 .

Jackson R. L., Busch S. J., and Cardin A. D. (1991) Glycosamino-glycans : molecular properties, protein interactions and role inphysiological processes . Physiol. Rev. 71, 481-539.

Ji Z., Fazio S., and Mahley R. W. (1994) Variable heparan sulfateproteoglycan binding of apolipoprotein E variants may modu-late the expression of type III hyperlipoproteinemia . J. Biol.Chem. 269, 13421-13428 .

Jin L., Ninomiya H., Roch J., Schubert D., Masliah E., Otero D. A.,and Saitoh T. (1994) Peptides containing the RERMS sequenceof amyloid,6 A4 protein precursor bind cell surface and promoteneurite extension . J. Neurosci. 14, 5461-5470.

Kang J., Lemaire H., Unterbeck A., Salbaum J., Masters C. L.,Grzeschik K., Multhaup G., Beyreuther K., and Muller-Hill B.(1987) The precursor of Alzheimer's disease amyloid A4 pro-tein resembles a cell surface receptor . Nature 325, 733-736.

Kitaguchi N., Takahashi Y., Tokushima Y., Shiojiri S., and Ito H.(1988) Novel precursor ofAlzheimer's disease amyloid precur-sor protein shows protease inhibitory activity . (Abstr .) Nature331,530-532.

Kjellen L. and Lindahl U. (1991) Proteoglycans : structures andinteractions (published erratum appears in Annu. Rev. Biochem.1992 ; 61 :following viii) . Annu . Rev. Biochem. 60, 443-475.

Konig G., Morning U., Czech C., Prior R., Banati R., Schreiter-Gasser U., Bauer J., Masters C., and Beyreuther K. (1992)Identification of a novel alternative splice isoform of the (JA4amyloid precursor protein (APP) mRNA in leukocytes andbrain microglial cells. J. Biol. Chem . 267, 10804-10809 .

Lohmander S., Shinomura T., Hascall V., and Kimura J. (1989)Xylosyl transfer to the core protein of the rat chondrosarcomaproteoglycan . J. Biol . Chem . 264, 18775-18780 .

Margolis R. K. and Margolis R. U. (1993) Nervous tissue proteogly-cans . Experientia 49, 429-446.

Mattson M. P., Cheng B ., Culwell A. R., Esch F. S., Lieberburg I .,and Rydel R. E. (1993) Evidence of excitoprotective effect andintraneuronal calcium-regulating roles for secreted forms of the0-amyloid precursor protein . Neuron 10, 243-254.

Mullan M., Crawford F., Axelman K., Houlden H., Lilius L., Win-blad B., and Lennfelt L. (1992) A pathogenic mutation forprobable Alzheimer's disease in the APP gene at the N-terminusof 6 amyloid . Nature Genet. 1, 345-347.

Murrell J., Farlow M., Ghetti B., and Benson M. D. (1991) Amutation in the amyloid precursor protein associated with he-reditary Alzheimer's disease . Science 254, 97-99.

Narindrasorasak S ., Lowery D., Gonzalez DeWhitt P., PoormanR. A., Greenberg B ., and Kisilevsky R. (1991) High affinityinteractions between the Alzheimer's beta-amyloid precursorproteins and the basement membrane form of heparan sulfateproteoglycan. J. Biol. Chem . 266, 12878-12883 .

Okamato M., Ichimura M., and Mori S. (1993) Excitotoxic deathof cultured cortical neurons : developmental changes and pro-

phylactic effect of chondroitin sulfate proteoglycans . Neurosci-ence 19, 163-172 .

Okamato M., Mori S ., Ichimura M., and Endo H . (1994) Chondroitinsulfate proteoglycans protect cultured rat's cortical and hippo-campal neurons from delayed cell death induced by excitatoryamino acids . Neurosci. Lett. 172, 51-54.

Pangalos M . N ., Efthimiopoulos S ., Shioi J ., and Robakis N . K .(1995) The chondroitin sulfate attachment site of appican isformed by splicing out exon 15 of the amyloid precursor gene.J. Biol. Chem . 270, 10388-10391 .

Prasad K . N . (1975) Differentiation of neuroblastoma cells in cul-ture . Biol . Rev. 50, 129--265 .

Refolo L ., Salton S ., Anderson J . P ., Mehta P ., and Robakis N. K.(1989) Nerve and epidermal growth factors induce the releaseof the Alzheimer amyloid precursor from PC-12 cell cultures .Biochem. Biophys. Res. Commun . 164, 664-670 .

Robakis N . K . (1994) (6-Amyloid and amyloid precursor protein, inAlzheimer's Disease (Terry R ., Kattman R., and Bicke E., eds),pp. 317-326 . Raven Press, New York .

Robakis N . K ., Ramakrishna N ., Wolfe G ., and Wisniewski H. M .(1987) Molecular cloning and characterisation of a cDNA en-coding the cerebral vascular and the neuritic plaque core amy-loid peptides. Proc. Natl. Acad. Sci . USA 84, 4190-4194 .

Robakis N . K., Altstiel L . D., Refolo L. M ., and Anderson J . P .(1990) Function and metabolism of the protease inhibitor con-taining Alzheimer amyloid precursors, in Molecular Biologyand Genetics ofAlzheimer's Disease (Miyatake T ., Selkoe D . J .,and Ihara Y ., eds), pp. 179-188 . Elsevier, New York.

Ruoslahti E. (1988) Structure and biology of proteoglycans . Annu .Rev . Cell Biol. 4, 229-255 .

Saitoh T ., Sundsmo H ., Roch J . M ., Kimura N ., Cole G ., SchubertD ., Oltersdorf T ., and Schenk D . B . (1989) Secreted forms ofamyloid /3 protein precursor is involved in the growth regulationof fibroblasts . Cell 58, 615-622 .

Sandbrink R ., Masters C . L., and Beyreuther K. (1994) Similaralternative splicing of a non-homologous domain in OA4-amy-loid protein precursor-like proteins . J. Biol. Chem . 269, 14227-14234 .

Schubert D ., LaCorbiere M., Saitoh T ., and Cole G. (1989a) Theregulation of amyloid 6 protein precursor and its modulatoryrole in cell adhesion . Neuron 3, 689-694 .

Schubert D., LaCorbiere M., Saitoh T ., and Cole G. (1989b) Charac-terisation ofan amyloid fl precursor protein which binds heparinand contains tyrosine sulfate . Proc. Natl. Acad. Sci. USA 85,2066-2069 .

Shioi J ., Anderson J. P ., Ripellino J. A ., and Robakis N . K. (1992)Chondroitin sulfate proteoglycan form of the Alzheimer's beta-amyloid precursor. J. Biol. Chem. 267, 13819-13822 .

Shioi J ., Refolo L. M ., Efthimiopoulos S ., and Robakis N . K. (1993)Chondroitin sulfate proteoglycan form of cellular and cell-sur-face Alzheimer amyloid precursor . Neurosci. Lett. 154, 121-124 .


Shioi J ., Pangalos M . N ., Ripellino J . A ., Vassilacopoulou D ., Mytili-neou C ., Margolis R. U., and Robakis N . K . (1995) The Alzhei-mer amyloid precursor proteoglycan (appican) is present inbrain and is produced by astrocytes but not by neurons in pri-mary neural cultures. J. Biol . Chem . (in press) .

Slunt H . H ., Thinakaran G., Von Koch C ., Lo A . C ., Tanzi R. E .,and Sisodia S . S . (1994) Expression of a ubiquitous, crossreactive homologue of the mouse 6-amyloid precursor protein(APP) . J. Biol. Chem . 269, 2637-2644.

Snow A. D ., Mar H., Nochlin D ., Kresse H ., and Wight T . N . (1992 )Peripheral distribution of dermatan sulfate proteoglycans (de-corin) in amyloid-containing plaques and their presence in neu-rofibrillary tangles of Alzheimer's disease . J. Histochem. Cyto-chem. 40, 105-113 .

Snow A. D ., Sekiguchi R., Nochlin D ., Fraser P., Kimata K ., Mizu-tani A., Arai M., Schreier W. A., and Morgan D . G . (1994a)An important role of heparan sulfate proteoglycan (perlecan)in a model system for the deposition and persistence of fibrillarA beta-amyloid in rat brain . Neuron 12, 219-234 .

Snow A . D ., Sekiguchi R . T ., Nochlin D ., Kalaria R. N., and KimataK . (1994b) Heparan sulfate proteoglycan in diffuse plaques ofhippocampus but not of cerebellum in Alzheimer's diseasebrain. Am. J. Pathol . 144, 337-347 .

Snow D . M ., Lemmon V., Carrino D . A ., Caplan A . I ., and SilverJ . (1990) Sulfated proteoglycans in astroglial barriers inhibitneurite outgrowth in vitro. Exp. Neural. 94, 767-776 .

Sprecher C . A ., Grant J. F ., Grimm G., O'Hara P . J ., Norris F .,Norris K ., and Foster D . C . (1993) Molecular cloning of thecDNA for a human amyloid precursor protein homolog : evi-dence for a multigene family . Biochemistry 32, 4481-4486 .

Su J . H ., Cummings B . J ., and Cotman C . W . (1992) Localizationof heparan sulfate glycosaminoglycan and proteoglycan coreprotein in aged brain and Alzheimer's disease . Neuroscience51, 801-813 .

Sunblad G ., Holojda S ., Roux L ., Varki A ., and Freeze H . H. (1988)Sulfated N-linked oligosaccharides in mammalian cells . II .Identification of glycosaminoglycan-like chains attached tocomplex-type glycans . J. Biol. Chem . 263, 8890-8896.

Tanzi R . E ., Gusella J. F ., Watkins P . C ., Burns G . A ., St. George-Hyslop P., Van Keuvan M . L ., Patterson D ., Pagan S ., KurnitD . M ., and Neve R. (1987) Amyloid beta protein gene : cDNAmRNA distribution and genetic linkage near the Alzheimer'slocus . Science 235, 880-884.

Thinakaran G . and Sisodia S . S . (1994) Amyloid precursor-likeprotein 2 (APLP2) is modified by the addition of chondroitinsulfate glycosaminoglycan at a single site . J. Biol. Chem. 269,22099-22104 .

Wasco W., Gurubhagavatula S ., Paradis M . D ., Romano D . M.,Sisodia S . S ., Hyman B . T ., Neve R. L ., and Tanzi R . E . (1993)Isolation and characterization of APLP2 encoding a homologueof the Alzheimer's associated amyloid,6 protein precursor . Na-ture Genet. 5, 95-100.


expression of the chondroitin sulfate proteoglycans of amyloid precursor (appican) and amyloid...

Documents