INTRODUCTION

Alzheimer's disease is a progressive dementia that is characterized by neuronal degeneration, synaptic loss, and the deposition of amyloid fibrils in the brain. The major constituent of the amyloid is the A protein (1, 2), which is derived from the amyloid precursor protein (APP).¹ APP has the structure of a transmembrane glycoprotein (3).

The function of APP is not clearly understood. APP expression is coordinately regulated with neuronal differentiation, neurite outgrowth, and synaptogenesis in the developing brain (4, 5, 6, 7). A number of studies have implicated APP in the regulation of neurite outgrowth in vitro (8, 9, 10, 11, 12, 13, 14, 15, 16).

The binding of APP to heparan sulfate proteoglycans (HSPGs), purified from postnatal day 3 mouse brain cells, has been shown to stimulate process outgrowth from central and peripheral neurons (17, 18, 19). Not all HSPG fractions were found to stimulate this action of APP. A preparation of HSPG from embryonic day 10 mouse brain cells did not stimulate APP-induced neurite outgrowth (17). This suggests that a specific, developmentally regulated HSPG may be responsible for activating the trophic function of APP.

A large number of developmentally expressed proteoglycans have been implicated in the regulation of neurite outgrowth. Proteoglycans are capable of providing signals that either stimulate or inhibit axonal growth in the developing nervous system (20). HSPG expression is generally associated with neuronal differentiation and synaptogenesis. The exact mechanism by which HSPGs facilitate neurite outgrowth is not clearly understood. Perlecan has been shown to have neurite outgrowth promoting activity in vitro (21). However, other HSPGs may stimulate neurite outgrowth by activating growth-enhancing proteins (22, 23, 24, 25) or by presenting growth factors to their cellular receptors (26, 27).

The aim of the present study was to identify and purify HSPGs that bind specifically to APP and to examine the effect of these HSPGs on the neurite outgrowth-promoting activity of APP. We show that the HSPG glypican (28) can bind strongly to APP and may regulate the neurite outgrowth-promoting properties of APP.

EXPERIMENTAL PROCEDURES

Dulbecco's modified Eagle's medium and Ham's F12 medium were purchased from ICN Biomedicals Australasia Pty. Ltd. (Seven Hills, Australia). IODO-GEN iodination reagent and BCA Protein Assay Reagent were purchased from Pierce. Sodium [³⁵S]sulfate (specific activity 100 mCi/mmol) was from NEN Research Products (North Ryde, Australia). Sodium [¹²⁵I]iodide and HyBond-N+ membranes were from Amersham Australia Pty. Ltd. (Castle Hill, Australia). Immobilon-P transfer membrane was from Millipore Corp. (Bedford, MA). Polylysine, leupeptin, aprotinin, pepstatin A, trans-epoxysuccinyl-L-leucylamido-(4-guanidino)butane (E-64), 4-(2-aminoethyl)benzenesulfonyl fluoride (AEBSF), putrescine, prostaglandin F2, progesterone, heparan sulfate from bovine kidney, Fast Red TR, naphthol AS-MX phosphate, 4-methylumbelliferyl--D-xyloside (-D-xyloside), and bovine serum albumin were purchased from Sigma-Aldrich Pty. Ltd. (Castle Hill, Australia). Econo-Pac Q cartridge (5 ml), Silver Stain Kit, and Affi-Gel 15 were purchased from Bio-Rad Laboratories Pty. Ltd. (North Ryde, Australia). Insulin-transferrin-selenium-X supplement (100 ×) was from Life Technologies, Inc. Heparitinase (heparin-sulfate lyase, EC 4.2.2.8) was purchased from Sapphire Bioscience (Alexandria, Australia). Laminin and perlecan were from Collaborative Biomedical Products (Bedford, MA). Cell culture plates were purchased from Nunc A/S (Kamstrup, Denmark). Nerve growth factor, chondroitinase ABC (EC 4.2.2.4), and CHAPS were from Boehringer Mannheim Australia Pty. Ltd. (Box Hill, Australia). Fetal calf serum and horse serum were obtained from Commonwealth Serum Laboratories Ltd. (Parkville, Australia). Dimethylmethylene blue was from Serva Feinbiochemica (Heidelberg, Germany). PD-10 columns, DEAE-Sepharose Fast Flow, and Sepharose CL-4B were purchased from AMRAD-Pharmacia Biotech (North Ryde, Australia). Rat monoclonal antibody 1948 which recognizes an epitope on the core protein of perlecan was purchased from Chemicon International Inc. (Temecula, CA). Rat monoclonal antibody 281-2 which recognizes an epitope on the core protein of mouse syndecan-1 (29) was a generous gift from Dr. Markku Jalkanen (University of Turku, Finland). Rabbit antisera MSE-2, MSE-3, and MSE-4 raised to the ectodomains of syndecan-2, syndecan-3, and syndecan-4 (30), respectively, were a generous gift from Dr. Merton Bernfield (Harvard Medical School, MA). Affinity-purified rabbit antisera 343-1 and 521-2² were raised to synthetic peptides homologous to amino acids 343-360 of rat glypican (31) and amino acids 521-535 of rat cerebroglycan (32), respectively. Goat anti-rabbit and anti-rat IgG conjugated to alkaline phosphatase were from Promega Corp. (Rozelle, Australia).

Human APP₆₉₅ was expressed in Pichia pastoris.³ Briefly, cDNA encoding APP₆₉₅ from leucine 18 to glutamine 611 was cloned into the expression vector pHIL-S1 and transfected into P. pastoris cells. Cells were grown in shaker flasks and APP₆₉₅ purified from the culture medium using Q-Sepharose, Mono-Q anion-exchange chromatography, and phenyl-Superose hydrophobic interaction chromatography (33).

Recombinant human APP was covalently coupled to Affi-Gel 15 by combining 4 mg of APP and 2 ml of gel in the presence of 100 mM Hepes buffer, pH 7.4. The gel slurry was gently mixed at 4 °C for 4 h. Active esters remaining on the gel were blocked by adding 200 µl of 1 M ethanolamine-HCl, pH 8.0, and gently mixing the gel for 1 h at 4 °C. The gel was poured into a 1 × 5-cm column and washed extensively with 150 mM NaCl, 20 mM Tris-HCl, pH 7.4 (Tris-buffered saline). Conditioned medium from cultures of dissociated postnatal day 3 mouse brain cells (19) was collected and centrifuged (1000 × g for 15 min at room temperature). A mixture of proteinase inhibitors was added to the medium (10 ml) to give a final concentration of 1 mM EDTA, 0.4 mM AEBSF, 2 µg/ml pepstatin A, 2 µg/ml aprotinin, 2 µg/ml leupeptin, and 5 µM E-64. The medium was applied to the APP affinity column which was washed with 20 ml of Tris-buffered saline. A linear salt gradient from 0 to 1.0 M NaCl in 20 mM Tris-HCl, pH 7.4, containing 0.02% (v/v) Tween 20 was applied to the APP affinity column over 30 ml at a flow rate of 1.0 ml/min and 1.0-ml fractions were collected. An aliquot (100 µl) from each fraction eluting from the APP affinity column was blotted onto HyBond-N+ membrane using a Bio-Dot Microfiltration Apparatus (Bio-Rad). Aliquots from each fraction were dialyzed against Tris-buffered saline containing 0.05% (v/v) Nonidet P-40 using a CF-35 Centriflo ultrafiltration membrane cone (Amicon, Danvers, MA) and then incubated with 50 milliunits/ml heparitinase in the presence of 1 mM calcium acetate, 0.4 mM AEBSF, 2 µg/ml pepstatin A, 2 µg/ml aprotinin, 2 µg/ml leupeptin, and 5 µM E-64 for 2 h at 37 °C. The samples were separated by sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) (34) containing 8.5% (w/v) acrylamide and then electrophoretically transfered to Immobilon-P transfer membrane. Both Immobilon-P and HyBond N+ membranes were incubated in 0.5% (w/v) hydrolyzed casein in phosphate-buffered saline (137 mM NaCl, 2.7 mM KCl, 10 mM Na₂HPO₄, 1.8 mM KH₂PO₄, pH 7.4) for 1 h. Membranes were then incubated with primary antibodies in Tris-buffered saline containing 0.05% (w/v) Tween 20 and 5% (v/v) goat serum. Antisera MSE-2, MSE-3, and MSE-4 were used at a dilution of 1:500; monoclonal antibodies 281-2 and 1948 and the affinity-purified 343-1 and 52--2 antisera were used at a concentration of 10 µg/ml. Anti-rabbit and anti-rat antibodies conjugated to alkaline phosphatase were used at 1 µg/ml in Tris-buffered saline containing 0.05% (w/v) Tween 20 and 5% (v/v) goat serum. The secondary antibody was visualized using Fast Red TR and naphthol AS-MX phosphate. The concentration of NaCl in the eluted fractions was determined by measuring conductivity in each fraction using an Activon model 301 conductivity meter (Activon, Carlton, Australia). Protein concentrations were determined with the BCA Protein Assay Reagent using bovine serum albumin as a standard.

HSPG was purified from the conditioned medium of rat pheochromocytoma (PC12) cells following a method for the purification of HSPG from the conditioned medium of human lung fibroblasts (35, 36). Rat pheochromocytoma (PC12) cells (37) were cultured for 4 days in Dulbecco's modified Eagle's medium supplemented with 10% (v/v) fetal calf serum and 5% (v/v) horse serum in 600-cm² cell culture dishes. Conditioned medium (1400 ml) was combined with medium (10 ml) from cells that had been labeled for 48 h with 200 µCi/ml of sodium [³⁵S]sulfate. Urea and Tris were added to the medium to a concentration of 6 M and 50 mM, respectively, and the final pH was adjusted to pH 8.0. The medium was applied to a column (2.5 × 20 cm) containing of DEAE-Sepharose Fast Flow (80 ml) at a flow rate of 2 ml/min. The column was washed extensively with a buffer containing 6 M urea, 1 mM EDTA, 0.1% (w/v) Triton X-100 in 20 mM Tris-HCl, pH 8.0, and bound protein was eluted with a gradient from 0 to 1.0 M NaCl over 800 ml in the same buffer. Fractions (10 ml) were collected. The amount of radioactivity in each fraction was determined using a 1600 TR Liquid Scintillation Analyzer (Packard International, Mt. Waverley, Australia). Fractions containing glypican were identified by dot blot and Western blot analysis using affinity-purified 343-1 antiserum. Glypican-containing fractions were pooled (120 ml), concentrated, and washed into Tris-buffered saline containing 0.05% (w/v) Nonidet P-40 using a CF-35 Centriflo ultrafiltration membrane cone (Amicon, Danvers MA). The sample was treated with chondroitinase ABC (100 milliunits/ml) for 3 h at 37 °C in the presence of 1 mM EDTA, 2 µg/ml pepstatin A, 0.4 mM AEBSF, 2 µg/ml leupeptin, and 5 µM E-64. The sample was then applied to a 5-ml Econo-Pac Q cartridge, and bound protein was eluted with a gradient from 0 to 1.0 M NaCl in 20 mM Tris-HCl, pH 7.4, containing 0.02% (w/v) Tween 20 over 30 ml. One-ml fractions were collected. Glypican-containing fractions were pooled (10 ml), concentrated, and washed into 4 M guanidine HCl, 1 mM EDTA, 1 mM benzamidine, 0.5% (w/v) CHAPS in 50 mM acetate buffer, pH 5.8, using a CF-35 Centriflo ultrafiltration membrane cone. The sample (600 µl) was applied to a column (1 × 100 cm) containing Sepharose CL-4B (78 ml) and eluted with the same buffer at a flow rate of 10 ml/h with 1-ml fractions collected. Glypican-containing fractions were pooled (16 ml) concentrated using a CF-35 Centriflo ultrafiltration membrane cone, and then the sample (600 µl) was desalted into 100 mM ammonium hydrogen carbonate, pH 8.0, using a PD-10 column. The sample was then lyophilized. Proteins were separated by SDS-PAGE containing 8.5% (w/v) acrylamide and protein bands visualized using a Coomassie Blue-silver stain method (38). The concentration of glycosaminoglycan was determined using dimethylmethylene blue (39) with heparan sulfate as a standard.

Purified glypican (1 µg) was iodinated using IODO-GEN iodination reagent (Pierce). Radiolabeled glypican was separated from free iodide using a PD-10 column. The specific radioactivity of the radiolabeled glypican was 2.9 × 10⁵ cpm/ng. Purified human brain APP (33) was added to a 96-well enzyme-linked immunosorbent assay plate (Greiner GmbH, Frickenhausen, Germany) at a concentration of 50 µg/ml in Tris-buffered saline; 100 µl was added to each well and incubated overnight at 4 °C. Nonspecific-binding sites were then blocked with 200 µl of 1% (w/v) bovine serum albumin in Tris-buffered saline for 2 h. Iodinated glypican was added to each well in 100 µl of 1% (w/v) bovine serum albumin in Tris-buffered saline for 1 h at room temperature. The wells were then washed five times with 150 µl of Tris-buffered saline containing 0.05% (w/v) Tween 20. Bound protein was then released from the well by adding 150 µl of 10% (w/v) SDS to each well and incubating at 60 °C for 30 min. The amount of radioactivity was determined using a 1261 Multigamma gamma counter (Wallac Oy, Turku, Finland). Specific binding was defined as the amount of glypican bound to a well incubated with APP minus the amount of glypican bound to a well incubated with blocking solution alone. Molar concentrations of glypican were determined by estimation of the protein concentration with the BCA Protein Assay Reagent with bovine serum albumin as a standard, assuming a molecular mass for the core protein of rat glypican of 62 kDa (40).

Cell culture plates (96-well) were prepared by incubating each well with 100 µl of 0.1 mg/ml polylysine in sterile distilled water for 15 min. The wells were then washed three times with 200 µl of sterile distilled water. The proteins of interest were then coated onto cell culture plates in order to be presented to the neurons in a substrate-bound form (substrate-bound). Wells were incubated with 10 µg/ml purified glypican or 10 µg/ml perlecan in 100 µl of phosphate-buffered saline for 2 h at room temperature. Each well was washed twice with 200 µl per well of phosphate-buffered saline. The wells were then incubated with 10 µg/ml of purified human brain APP or 5 µg/ml laminin in 100 µl of phosphate-buffered saline for 2 h and then wells were washed twice with 200 µl of phosphate-buffered saline. The order in which the proteins were added to the wells did not affect the results. Approximately 2 × 10³ cells, prepared from embryonic day 12 chick paravertebral sympathetic ganglia (41), were added to each well in 100 µl of a 1:1 mixture of Dulbecco's modified Eagle's medium and Ham's F12 media containing 1.0% (v/v) Insulin-transferrin-selenium-X supplement (100 ×) 100 µM putrescine, 1.67 µg/ml prostaglandin F2, 6.67 ng/ml progesterone, 10 ng/ml NGF, and 0.5 mg/ml bovine serum albumin. In some experiments, cultures were allowed to grow for 1 h after plating, and then the medium was removed and replaced with fresh medium containing 1 mM -D-xyloside. Cultures were grown for 48 h, fixed with 2% (w/v) glutaraldehyde in phosphate-buffered saline, and then examined by phase-contrast microscopy. Selected fields were captured for computer-assisted image analysis (MD30 Plus Image analysis system, Adelaide, Australia). The percentage of neurons with neurites was determined in nine separate fields around the center of each well. On average 60 neurons were counted per well. For the analysis of neurite length, the longest neurite on 20 neurons bearing neurites in each well was measured. A minimum of four wells for each treatment group was analyzed. Differences between the means of controls and each treatment group were analyzed by a two-tailed Student's t test. Means were assumed to be significantly different when the p value for the null hypothesis was less than 0.05.

RESULTS

An affinity column containing recombinant APP₆₉₅ as a ligand was used to purify HSPGs from neonatal mouse brain which bind strongly to APP. Conditioned medium from postnatal day 3 mouse brain cultures was applied to the APP affinity column, and bound protein was eluted with a linear salt gradient (Fig. 1A). Aliquots of each fraction eluted from the column were assayed by Western blotting and dot blotting using antibodies to a number of well characterized HSPGs. The antibodies tested were specific to syndecan-1, syndecan-2, syndecan-3, and syndecan-4, perlecan, cerebroglycan, and glypican. Western blot and dot blot assays using antibodies to syndecan-1, syndecan-2, syndecan-3, syndecan-4, perlecan, or cerebroglycan did not detect any immunoreactivity in fractions eluted from the column. However, antibodies to glypican (affinity-purified 343-1 antiserum) detected immunoreactivity eluting between fractions 15 and 25 (Fig. 1A). The peak of 343-1 immunoreactivity eluted at 0.39 M NaCl (fraction 20).

Western blot analysis of the eluate revealed a diffuse immunoreactive band migrating with an apparent molecular mass of 200 kDa (Fig. 1B, lane 1). Pretreatment of the eluate with heparitinase resulted in the loss of the 200-kDa band and the appearance of an immunoreactive band with an apparent molecular mass between 60 and 67 kDa (Fig. 1B, lane 2), consistent with the apparent molecular mass of the core protein of rat glypican (31).

To prepare a sufficient amount of glypican for binding and cell culture studies, glypican was purified from the conditioned medium of rat pheochromocytoma (PC12) cells. The conditioned medium of the PC12 cells contains a soluble form of glypican which has probably been cleaved from the cell-membrane adjacent to its glycosylphosphatidylinositol (GPI) anchor (42). PC12 cells were cultured in the presence of [³⁵S]sulfate to radiolabel proteoglycans. DEAE-Sepharose Fast Flow chromatography of PC12-conditioned medium was used as a first step to purify glypican (Fig. 2A). The amount of radioactivity in each fraction was determined. The presence of glypican in the fractions was determined by dot blot and Western blot analysis using the affinity-purified 343-1 antiserum. Glypican eluted between 0.25 and 0.35 M NaCl.

To remove chondroitin sulfate proteoglycans, glypican-containing fractions (fractions 32-46 from the DEAE-Sepharose Fast Flow column) were pooled and treated with 100 milliunits/ml chondroitinase ABC and then applied to an Econo-Pac Q cartridge (Bio-Rad). Glypican eluted between 0.6 and 0.8 M NaCl (Fig. 2B). Glypican-containing fractions (fractions 28-39 from the Econo-Pac Q cartridge) were pooled and concentrated and then applied to a Sepharose CL-4B gel filtration column (1 × 100 cm). Glypican eluted between fractions 50 and 65 (Fig. 2C). Glypican-containing fractions from the Sepharose CL-4B column were pooled, concentrated, desalted, and lyophilized.

Purified glypican was analyzed by SDS-PAGE and Western blot. Glypican did not stain strongly using a Coomassie Blue-silver stain which may be due to its high level of glycosylation and the diffuse nature of the band. Heparitinase treatment of glypican generated two bands that were not observed in the heparitinase preparation alone, with one band migrating at 64 kDa and the other migrating at 50 kDa (Fig. 3A). Western blot analysis of the purified glypican with the 343-1 antibodies revealed a high molecular weight immunoreactive band migrating between 140 and 220 kDa. Pretreatment of the purified glypican with heparitinase, or a combination of heparitinase and chondroitinase ABC, resulted in the loss of the 140- to 220-kDa band and the appearance of an immunoreactive band between 58 and 70 kDa and an immunoreactive band at 49 kDa (Fig. 3B). The 49-kDa band detected by Western blot analysis and the 50-kDa band detected by the Coomassie Silver stain may represent a proteolytic breakdown product of the core protein of glypican. The recovery of glypican from the conditioned medium was 0.18 µg of bovine serum albumin eq/ml and 0.07 µg of heparan sulfate eq/ml.

The binding of APP to purified glypican was examined using a solid-phase binding assay. A range of glypican concentrations (0 to 4.0 nM) was added to the APP-coated enzyme-linked immunosorbent assay plates. Binding of glypican to APP was saturable, and Scatchard analysis of the binding indicated one class of binding site with an equilibrium dissociation constant (K_D) of 2.8 nM (Fig. 4).

Previous studies (19) have shown that substrate-bound APP can stimulate neurite outgrowth in the absence of exogenously added HSPG, provided that cells are cultured for 48 h or longer in serum-free medium. The effect of glypican on neurite outgrowth from sympathetic neurons cultured on substrate-bound APP and laminin was examined (Fig. 5). Cultures were grown for 48 h, and the percentage of neurons with neurites and neurite length were determined (Table I). Laminin and polylysine induced a 25-fold increase in the number of neurons with neurites and an 11-fold increase in neurite length compared with neurons cultured on polylysine alone. Neurons cultured on APP and polylysine exhibited a 10-fold increase in the number of neurons with neurites and a 5-fold increase in neurite length compared with neurons cultured on polylysine alone. Glypican inhibited APP-stimulated neurite outgrowth 76%, as measured by the number of neurons with neurites, and 62%, as measured by neurite length. Glypican did not significantly affect the number of neurons with neurites or neurite length of neurons cultured on laminin. Neurons cultured on polylysine and glypican did not exhibit any increase in both the number of neurons with neurites or neurite length above that observed on the polylysine alone.

Phase-contrast micrographs of primary cultures of embryonic day 12 chick sympathetic neurons cultured for 48 h on substratum-bound APP and laminin in the presence or absence of glypican or in the presence and absence of 1 mM -D-xyloside.

Table I. Effect of glypican on neurite outgrowth from sympathetic neurons cultured on substrate-bound APP and laminin Neurons were cultured for 48 h on cell culture plates coated with 10 µg/mL APP and 5 µg/ml Laminin with and without coating with 10 µg/ml glypican or 10 µg/ml perlecan. Computer-assisted image analysis (MD30 Plus Image) was performed on the cultures. The percentage of neurons with neurites was determined in nine separate fields around the center of each well. The length of the longest neurite on 20 neurons was measured. A minimum of four wells for each treatment group was analyzed. Data represent the mean and standard error of the mean. Substrate Percentage of neurons with neurites Neurite length µm Polylysine 3.6  ± 3.0 22  ± 2.7 Polylysine + glypican 2.3  ± 1.3 19  ± 4.4 Polylysine  APP 37  ± 6.0a 114  ± 10a Polylysine + glypican + APP 9.0  ± 2.2b 43  ± 10b Polylysine + laminin 90  ± 15 247  ± 37 Polylysine + glypican + laminin 92  ± 7.2 223  ± 25 Polylysine + perlecan + APP 10.7  ± 2.7b 49.5  ± 9.0b Polylysine + perlecan + laminin 95  ± 6.3 214  ± 18 a  Significantly different from polylysine (p < 0.05). b  Significantly different from polylysine + APP (p < 0.05).

The effect of glypican on APP- and laminin-induced neurite outgrowth was compared with the effect of the basement membrane HSPG, perlecan. Perlecan inhibited APP-stimulated neurite outgrowth 71%, as measured by the number of neurons with neurites, and 56%, as measured neurite length. Perlecan did not significantly affect the number of neurons with neurites or neurite length of neurons cultured on laminin.

Effect of -D-Xyloside on APP-induced Neurite Outgrowth

The effect of inhibiting endogenous proteoglycan on APP- and laminin-induced neurite outgrowth was investigated. Cultures were grown in the presence or absence of 1 mM -D-xyloside (Fig. 5), a competitive inhibitor of proteoglycan glycanation (43). The presence of 1 mM -D-xyloside resulted in an 80% inhibition of proteoglycan glycanation as determined from the incorporation of [³⁵S]sulfate in secreted and cell-associated proteoglycans (Table II). The percentage of neurons with neurites and neurite length were determined (Table III). Neurons cultured on substrate-bound APP in the presence of -D-xyloside exhibited an 88% reduction in the number of neurons with neurites and a 79% reduction in neurite length compared with neurons cultured in the absence of -D-xyloside. The presence of -D-xyloside did not affect the percentage of neurons with neurites and did not affect neurite length of neurons cultured on substrate-bound laminin compared with neurons cultured in the absence of -D-xyloside.

Table II. The effect of -D-xyloside on the incorporation of [35S]sulfate into proteoglycans and glycosaminoglycans produced by sympathetic neurons in culture Sympathetic neurons were grown on cell culture plates coated with 10 µg/ml APP and labeled with 400 µCi/ml of [35S]sulfate for 48 h in the presence and absence of 1 mM 4 methylumbelliferyl--D-xyloside. The incorporation of [35S]sulfate into low molecular weight glycosaminoglycan and high molecular weight proteoglycans was determined by size exclusion high performance liquid chromatography (5). Data represent the mean and standard error of the mean (n = 3). Fraction Incorporation of [35S]sulfate (cpm/103 cells) Control  -D-Xyloside (1 mM) Control (%) Medium   Proteoglycans 288,500  ± 31,800 59,400  ± 7,200 21   Glycosaminoglycans 45,800  ± 7,200 1,448,100  ± 120,000 3,162 Cells   Proteoglycans 40,400  ± 6,000 7,000  ± 2,700 17   Glycosaminoglycans 50,500  ± 8,200 68,700  ± 7,800 136

Table III. Effect of xyloside on neurite outgrowth from sympathetic neurons cultured on substratum-bound APP and laminin Neurons were cultured for 48 h in the presence or absence of 1 mM 4-methylumbelliferyl--D-xyloside (-D-xyloside). Computer-assisted image analysis (MD30 Plus Image) was performed on the cultures. The percentage of neurons with neurites was determined in nine separate fields around the center of each well. The length of the longest neurite on 20 neurons was measured. A minimum of four wells for each treatment group was analyzed. Data represent the mean and standard error of the mean. Treatment Percentage of neurons with neurites Neurite length µm Polylysine 7.6  ± 2.1 29  ± 8.4 Polylysine + -D-xyloside 5.8  ± 2.0 26  ± 9.0 Polylysine + APP 45  ± 5.0 119  ± 9.2 Polylysine + APP + -D-xyloside 5.3  ± 1.9a 25  ± 7.0a Polylysine + laminin 92  ± 2.4 209  ± 13 Polylysine + laminin + -D-xyloside 94  ± 2.2 200  ± 19 a  Significantly different from polylysine + APP (p < 0.05).

DISCUSSION

There is increasing evidence that the physiological effects of secreted APP are mediated through an interaction with components of the extracellular matrix such as HSPGs. HSPGs constitute a majority of binding sites for APP in the extracellular matrix (5) and up to 50% of the binding sites for APP on the cell surface (13). Previous studies have shown that specific, developmentally regulated HSPGs can modulate the neurite outgrowth-promoting activity of APP (17, 18). In this study, affinity chromatography using recombinant APP as a ligand showed that the HSPG glypican can bind to APP. The equilibrium dissociation constant for this interaction was low (2.8 nM) suggesting that this is a strong interaction of biological significance. The equilibrium dissociation constant for the binding of APP to glypican was similar to that reported for the binding of APP to the basement membrane form of HSPG (perlecan) (44). Perlecan was not detected in the eluate from the APP affinity column. However, this is likely to have been due to the low expression of perlecan in the cultured mouse brain cells, rather than to any difference in the relative affinities of glypican and perlecan for APP during affinity chromatography.

Cell culture studies demonstrated that exogenous glypican can specifically inhibit the neurite outgrowth-promoting activity of APP. Glypican did not inhibit neurite outgrowth on laminin, indicating that this effect may be specific for APP. In addition, neurite outgrowth on substrate-bound APP was significantly reduced by inhibiting proteoglycan glycanation with -D-xyloside. However, -D-xyloside did not significantly reduce neurite outgrowth on laminin. Taken together, these results suggest that endogenous proteoglycans either on the cell surface, or secreted by the neuron, are required for the promotion of neurite outgrowth on substrate-bound APP. Therefore, exogenous glypican may inhibit APP-induced neurite outgrowth by competing with endogenous proteoglycans for binding to APP. This view is supported by the inhibitory effect of perlecan on APP-induced neurite outgrowth. APP has been shown to bind to perlecan with high affinity (44) which suggests that perlecan, like exogenous glypican, inhibits APP-stimulated neurite outgrowth by competing with endogenous proteoglycans for binding to APP. The endogenous proteoglycan remains to be identified; it could be GPI-anchored glypican. However, other cell-surface or secreted proteoglycans could also be involved.

The results presented in this study provide further evidence of the importance of proteoglycans in APP-stimulated neurite outgrowth. The present study suggests that glypican may be involved in this mechanism. Glypican is found either as a GPI-anchored cell-surface HSPG or, when cleaved from its GPI anchor, a soluble HSPG (28). The glypican used in this study would be the secreted form as it was purified from conditioned medium. Glypican is found in high levels in the brain where it is primarily expressed by neurons (31, 45). The view that glypican may be involved in neurite outgrowth is consistent with a recent study that has shown that glypican is expressed on the surface of most axons during periods of major axon growth.²

The precise mechanism by which APP promotes neurite outgrowth remains to be elucidated. The neurite outgrowth-promoting domain of APP has been associated with the RERMS sequence contained within the ectodomain of the protein (14). However, other studies place the neurite outgrowth-promoting domain closer to the carboxyl terminus of the ectodomain of APP (16), a region of APP implicated in modulating intracellular calcium levels (46). A peptide homologous to the RERMS-containing domain of APP has been shown to stimulate neurite outgrowth in a manner that is independent of heparan sulfate binding (13). However, the present study suggests that full-length APP requires the presence of endogenous proteoglycans in order to promote neurite outgrowth.

A number of proteoglycans, including perlecan, have been localized to the amyloid plaques of Alzheimer's disease (47, 48, 49, 50). Little is known about the effects these proteoglycans have on the function and metabolism of APP which may be secreted in the region of the amyloid plaques. As proteoglycans have been shown to modulate the neurite outgrowth-promoting activity of APP, they may perturb the normal function of APP in the region of the amyloid plaque. What significance, if any, such a perturbation may have for the pathogenesis of Alzheimer's disease remains to be determined.