Colocalization and Complex Formation ... - Wiley Online Library

Journal ofNeurochemistry Lippincott—Raven Publishers, Philadelphia © 1998 International Society for Neurochemistry

Colocalization and Complex Formation Between Prosaposin and Monosialoganglioside GM3 in Neural Cells R. Misasi, M. Sorice, T. Garofalo, T. Griggi, *W. M. Campana, tM. Giammatteo, ~A. Pavan, *M. Hiraiwa, G. M. Pontieri, and *J~S. O’Brien Dipartimento di Medicina Sperimentale e Patologia, Università “La Sapienza,” Roma; tCentro di Microscopia, ~Dipartimento di Medicina Sperimentale, Università deli’ Aquila, Aquila, Italy; and * Department of Neurosciences and Center for Molecular Genetics, University of Cai~fornia,San Diego, School of Medicine, La Jolla, California, U.S.A.

Abstract: Prosaposin, the precursor of saposins A, B, C, and D, was recently identified as a neurotrophic factor in vitro as well as in vivo. Its neurotrophic activity has been localized to a linear 12-amino acid sequence located in the NH2-terminal portion of the saposin C domain. In this study, we show the colocalization of prosaposin and ganglioside GM3 on NS2OY cell plasma membrane by scanning confocal microscopy. Also, TLC and western blot analyses showed that GM3 was specifically associated with2~prosaposin -independent in immunoprecipitates; and not disassociatedthis during binding sodium was Ca dodecyl sulfate—polyacrylamide gel electrophoresis. The association of prosaposin—GM3 complexes on the cell surface appeared to be functionally important, as determined by differentiation assays. Neurite sprouting, induced by GM3, was inhibited by antibodies raised against a 22-mer peptide, prosaptide 769, containing the neurotrophic sequence of prosaposin. In addition, pertussis toxin inhibited prosaptide-induced neurite outgrowth, as well as prosaptide-enhanced ganglioside concentrations in NS2OY cells, suggesting that prosaposin acted via a G protein-mediated pathway, affecting both ganglioside content and neuronal differentiation. Our findings revealed a direct and tight GM3—prosaposin association on NS2OY plasma membranes. We suggest that ganglipside—protein complexes are structural components of the prosaposin receptor involved in cell differentiation. Key Words: Prosaposin —Gangliosides—GM3— NS2OY cells—Neural differentiation. J. Neurochem. 71, 2313—2321 (1998).

Prosaposin is the precursor of saposins A, B, C, and D, which activate lysosomal hydrolysis of sphingolipids (O’Brien and Kishimoto, 1991). Partially proteolyzed products derived from prosaposin have been isolated and identified (Hiraiwa et al., 1997a,b). Prosaposin exists as a secretory protein in human milk, CSF, and seminal plasma (Hiraiwa et a!., 1993), but it is also present at high concentration in human and rat brain and occurs as an integral membrane component of neuronal and other plasma membranes (Rijnboutt 2313

et a!., 1991; Fu et al., 1994). Northern analysis of prosaposin mRNA during embryonic development showed a very high concentration in both mouse brain and dorsal root ganglia, indicative of a developmental role of prosaposin (Levy et al., 1991). Prosaposin bound to a putative high-affinity receptor that induced protein phosphorylation (O’Brien et al., 1994) and was associated with a G protein, Goa (Hiraiwa et al., 1997a,b). The binding between prosaposin and receptor triggered a signal transduction cascade, with induction of extracellular signal-regulated protein kinase phosphorylation in neural and glial cells (Campana et al., 1996; Hiraiwa et al., l997a,b). In primary Schwann cells prosaposin activated the mitogen-activated protein kinase cascade through a pertussis toxinsensitive G protein-mediated pathway (Campana et al., 1998). In addition, it was demonstratedthat prosaposin is a neurotrophic factor in vitro (O’Brien et al., 1994) as well as in vivo (Kotani et al., 1996), and its neurotrophic activity has been localized within a linear 12mer residue sequence located in the NH 2-terminal portion of the saposin C domain (O’Brien et al., 1995). Several active synthetic peptides (14—22 residues) encompassing this region, named prosaptides, have been synthesized (O’Brien et al., 1995; Qi et a!., 1996). The neurotrophic activity of prosaposin in NS2OY neuroblastoma cells might be mediated in part by the increase in content of cell surface gangliosides. We Received April 27, 1998; revised manuscript received July 17, 1998; accepted July 17, 1998. Address correspondence and reprint requests to Dr. J. S. O’Brien at Department of Neurosciences, University of California, San Diego, School ofMedicine, 9500 Gilman Drive, La Jolla, CA 920930634, U.S.A. Abbreviations used: anti-769 Ab, antibody raised against the active 22-mer prosaptide 769; CT, cholera toxin; DMEM, Dulbecco’s modified Eagle’s medium; FITC, fluorescein isothiocyanate; HPTLC, high-performance TLC; M anti-saposin C Ab, mouse monoclonal anti-human saposin C; mAb, monoclonal antibody; NGF, nerve growth factor; R anti-saposin C Ab, rabbit polyclonal anti-human saposin C; PBS, phosphate-buffered saline; TRITC, tetramethyirhodamine B isothiocyanate.

2314

R. MISASI ET AL.

recently demonstrated that prosaposin, as well as a 22mer prosaptide, prosaptide 769, induced a significant increase of ganglioside concentration in NS2OY neuroblastoma cell plasma membranes (Misasi et al., 1996). Modulation of membrane lipid concentration and, consequently, lipid—protein interactions has been considered as a possible mechanism for controlling neuronal differentiation (Schengrund, 1990). Indeed, glyco!ipids, especially the monosia!ogangliosides GMI and GM3, have been shown to exert their biological functions in addition to or in association with specific membrane proteins, such as several growth factor receptors (Bremer, 1994). GM3 was able to down-regulate fibroblast growth factor and epidermal growth factor receptor activities through inhibition of tyrosine kinase activity (Bremer and Hakomori, 1982; Bremer et a!., 1986), whereas GM1 specifically inhibited plateletderived growth factor receptor activity (Bremer et al., 1984) and enhanced the action of nerve growth factor (NGF) by binding to Trk A, the tyrosine kinase receptor for NGF (Mutoh et a!., 1995). Moreover, gangliosides may control cell growth and differentiation by modulating protein kinase C activity (Kreutter et a!., 1987) and inhibiting insulin receptor-associated kinase activity (Bremer et a!., 1984). Recently, gangliosides have been identified as inhibitors of ADP-ribosy!transferases (Hara-Yokoyama et a!., 1995), which cause the functional uncoupling of G proteins from receptors (Sullivan et a!., 1987), Suggesting that gang!iosides may play a role in regulating G protein activity. Prosaposin has been shown to bind gangliosides with high affinity and to facilitatetheir transfer from liposomes to biological membranes (Hiraiwa et a!., 1992) in vitro. In the present study we showed that prosaposin cobcalized with GM3 and formed a tight complex in vitro, indicating that this interaction may play a role in neurite outgrowth. EXPERIMENTAL PROCEDURES Cells Murine NS2OY (cholinergic) neuroblastoma cells, which extend neurites in response to gangliosides (Uemura et al.,

199!) were from Prof. K. Uemura and T. Taketomi. Cells were maintained in Dulbecco’s modified Eagle’s medium (DMEM; Bio Whittaker, Walkersvi!le, MD, U.S.A.), containing 10% fetal calf serum plus 100 units /ml penicillin, 100 ~tg/ml streptomycin, and 250 pg/ml amphotericin B (Fungizone) at 37°Cunder a humidified 5% CO2 atmosphere. The cells were serially passaged After 0.25% ‘—~24h, trypsin the 2 flasks.with at a splitwas medium ratioreplaced of 1:20 with in 25-cm DMEM containing 0.5% fetal calf serum plus effectors. Cells were cultured for an additional 24 h and analyzed for ganglioside pattern and neurite outgrowth.

Effectors Prosaposin was prepared as previously described from

human milk (Hiraiwa eta!., 1993). Synthetic prosaptide was obtained commercially (Peptide Synthesis Facility, University of California, San Diego). The peptide was purified by

J. Neurochem., Vol. 71, No. 6, 1998

HPLC on a Vydac C-4 column to >95% purity before use. Prosaptide 769, as described above, is a 22-mer peptide,

which is active as a neurotrophic factor (CEFLVKEVTKLIDNNKTEKEIL) (O’Brien eta!., 1995). In all the experiments cells were stimulated with prosaptide 769 at 100 ng/ ml for 24 h. Monosialoganglioside GM3 was from Fidia Research Laboratory (Padua, Italy).

Antibodies A rabbit polyclonal anti-human saposin C (R anti-saposin C Ab), reacting with saposin C and prosaposin but not with gangliosides, as judged by electrophoretic techniques and ELISA, was prepared as previously described (Morimoto et al., 1988) and purified by affinity chromatography using saposin-AffiGel conjugates (Hiraiwa eta!., !997a,b). Affinity matrixes were prepared by coupling of saposin C (10 mg) with AffiGel 10(2 ml of gel) according to the protocol recomn-iended by the manufacturer. This antibody was used for all the immunoprecipitations. A mouse monoclonal anti-human saposin C (M anti-saposin C Ab), reacting with saposin C and prosaposin but not with gangliosides, as judged by electrophoretic techniques and ELISA, was prepared as previously described (Stastny et al., 1992) and also purified from the culture media by affinity chromatography using protein G-Sepharose (Hiraiwa et a!., 1997a,b). This antibody was used for all western blot analyses. An antibody was raised against the active 22-mer prosaptide 769 (anti-769 Ab) by immunizing rabbits with the pep-

tide conjugated to keyhole limpet hemocyanin (Chiron Mimotopes, San Diego, CA, U.S.A.) (O’Brien et al., 1995). This antibody was used for a!! immunofluorescence experiments. Anti-GM3 (GMR6), anti-GM2 (GMB28), and anti-GM1 (GMB16) (Kotani et al., 1992) monocbonal antibodies (mAbs) were kindly provided by Dr. T. Tai (Tokyo Metropolitan Institute of Medical Science, Tokyo, Japan). These antibodies did not bind directly to prosaposin, as revealed

by ELISA and western blotting. Immunofluorescence staining Cells (1 x !0~in 1 ml of DMEM containing 10% fetal calf serum) were immobilized by settling onto coverslips that had been coated with 300 ~.tlof 2% pig skin gelatin .(Sigma Chemical Co., St. Louis, MO, U.S.A.) in phosphatebuffered saline (PBS; pH 7.4) and then incubated overnight at 37°Cin a humidified atmosphere of 5% CO 2. After washing three times with PBS, cells were fixed in 4% formaldehyde in PBS for 1 h at 4°C.After washing three times in PBS, cells were incubated for 1 h at 4°C with rhodamine [tetramethylrhodamine B isothiocyanate (TRITC ) I -conjugated cholera toxin (CT) B subunit for a direct staining or with anti-GM3 (GMR6) or with anti-GM2 (GMB28), followed by addition (30 mm at 4°C)of goat anti-mouse

1gM (~.tchain-specific) conjugated with Texas Red (Calbiochem, La Jolla, CA, U.S.A.), for an indirect staining. After washing with PBS, cells were incubated for 1 h at 4°Cwith anti-769 Ab, diluted 1:20 in PBS, followed by addition of fluorescein isothiocyanate (FITC ) -conjugated goat anti-rabbit IgG (y chain-specific; Sigma) and incubated for an additional 30 mm at 4°C.Separately, in parallel experiments, cells were directly stained with rhodamine (TRITC) -conjugated CT B subunit, anti-GM3 (GMR6), or anti-GM2 (GMB28) before fixing the cells. Alternatively, cells were processed for a second formaldehyde fixation immediately

2315

PROSAPOSIN-GM3 COMPLEXES IN NEURAL CELLS after the incubation with anti-GM3 mAb and before the secondary Texas Red anti-mouse 1gM. Cells were finally washed three times in PBS and then mounted upside down onto a glass slide in 5 btl of g!ycerol/Tris-HC1, pH 9.2. The

covers!ips were sealed with nail varnish to prevent evaporation and stored at 4°C before imaging. The images were acquired through a confocal laser scanning microscope (Sarastro 2000; Molecular Dynamics) equipped with a Nikon Optiphot microscope (objective 60/1.4 oil) and an argon ion laser (25 mW output). Simultaneously, the green (FITC) and the red (TRITC for the CT series, Texas Red for the anti-GM3 and anti-GM2 series) fluorophores were excited at 488 and 518 nm and were observed by two different detectors. Before FITC-TRITC image acquisition, the samples were scanned at different filter conditions to choose a setting that reduced overlap of emission spectra with a maximal signal-to-noise ratio. Then, acquired images were processed by subtracting a scaled version of green from red series, and vice versa. The scale factor (bleed-through factor) was determined by scanning single-stained samples in

a dual fluorescence scanning configuration. For anti-GM3— anti-prosaposin and anti-GM2 anti-prosaposin dual fluorescence, samples were stained with Texas Red fluorophore, —

which greatly reduces fluorescence overlapping. Images were collected at 512 X 512 pixels (0.17 ~m per pixel lateral dimension, 0.6 ~m per pixel axial dimension). Serial optical sections were assembled in depth-coding mode. Acquisition and processing were carried out using Image Space software (Molecular Dynamics).

Immunoprecipitation and immunoblotting analysis GM3-treated NS2OY neuroblastoma cells (10 ~zg/mI per 106 cells), with or without a 15-mm pretreatment with 10 mM EGTA, were lysed in lysis buffer as reported above. Cell-free lysates were normalized for protein content, immunoprecipitated as described above, using the same antibody (R anti-saposin C Ab), electrophoretically transferred to nitrocellulose membrane (Bio-Rad, Hercules, CA, U.S.A.) after sodium dodecyl sulfate—polyacrylamide gel electrophoresis with 10% polyacrylamide gels, and then probed with anti-GM3 (GMR6), anti-GM2 (GMB28), or anti-GM! (GMB16) mAb, diluted 1:20 in PBS. Bound antibodies were then visualized with anti-mouse 1gM horseradish peroxidaseconjugated secondary antibody diluted 1:500 in PBS, followed by incubation with the enhanced chemiluminescence western blotting detection reagent (Amersham Corp., Buckinghamshire, U.K.). Manufacturer-specified protocols were used to strip anti-GM3 antibody from the blots (ECL manual, Amersham) and to reprobe the membranes with M antisaposin C Ab, which recognized prosaposin as well, followed by addition of horseradish peroxidase-conjugated goat anti-mouse IgG (y chain-specific; Sigma) diluted 1:5,000 in PBS.

Analysis of cell differentiation Cell differentiation was evaluated by determining neurite outgrowth in NS2OY cells assayed by the method described by Uemura et al. (1991). In particular, cells (2 x 10k) were

plated onto glass coverslips in 30-mm-diameter Petri dishes. Immunoprecipitation and high-performance TLC (HPTLC) analysis of ganglioside extracts NS2OY neuroblastoma cells were lysed in lysis buffer [20 mM HEPES (pH 7.2), 1% Nonidet P-40, 10% glycerol, 50 mM NaF, 1 mM phenylmethylsulfonyl fluoride, 1 mM Na 3VO4, and 10 ,ag of leupeptin/ml I. Cell-free lysates were normalized for protein content and immunoprecipitated with R anti-saposin C Ab, which binds prosaposin with high affinity. Gangliosides present in the prosaposin precipitates were extracted according to the method of Svennerholm and Fredman (1980), with minor modifications. In brief, cells were extracted twice in chloroform/methanol/water (4:8:3 by volume) and subjected to Fo!ch partition by addition Of water to give a final chloroform/methanol/water ratio of 1:2:1.4. The upper phase, containing polar glycosphingolipids, was desalted, and low-molecular-weight contaminants were removed using Supelclean LC-!8 tubes (Supelco, Bellefonte, PA, U.S.A.), according to the method of Williams and McC!uer (1980). The eluted glycosphingolipids were dried and separated by HPTLC, using silica gel 60 HPTLC plates (Merck, Darmstadt, Germany). Chromatography was performed in chloroformlmethanol/0.25% aqueous KC1 (5:4:1 by volume).

Ganglioside standards GM3 (FIDIA), GMI, GD1a, GD1b, and GT1b (Sigma) were included in every HPTLC plate analysis. Plates were air-dried, and gangliosides were visual-

ized with resorcinol, which specifically stains sialic acidcontaining glycosphingolipids (Svennerholm, 1957). TLC immunostaining was performed on aluminumbacked silica gel plates (Merck), as previously described (Misasi eta!., 1993), using the anti-GM3 mAb GMR6 (Kotani et a!., 1992).

After 24 h, the medium was replaced with DMEM containing 0.5% fetal calf serum plus effectors: (a) GM3 (100 ~.tg/ml), in the absence or in the presence of anti-769 Ab (2.5 ~.tl/ ml) and, as a control, of preimmune serum; or (b) prosaposin (!nM) or prosaptide 769 (1 nM), with or without pertussis toxin (100 ng/ml; Calbiochem), as a metabolic inhibitor, 4 h before addition of effector. Cells were cultured for 24 h and analyzed for ganglioside pattern and neurite outgrowth. Neurite outgrowth was scored under a phase-contrast microscope. Cells bearing neurites longer than one cell diameter were scored as positive, and 100 cells were counted in triplicate from different portions of each dish. Each assay was carried out in duplicate dishes. Four independent experiments were performed.

Ganglioside incorporation and extraction NS2OY neuroblastoma cells were incubated with GM! or GM3 (100 ~tg/ml) in the presence of anti-769 Ab diluted 1:20 in PBS. NS2OY neuroblastoma cells were incubated with GM3 (100 ~sg/ml per io~cells) in the presence or in

the absence of anti-769 Ab diluted 1:20 in PBS. Alternatively, cells cultured in medium with 0.5% fetal calf serum were incubated for 24 h with prosaposin (1 nM) or prosaptide 769 (100 ng/ml), in the presence or in the absence of pertussis toxin (100 ng/ml). Cells were washed with PBS containing 10% fetal calf serum and PBS containing 0.1% trypsin to remove loosely associated gangliosides (Chigorno et al., 1985). Gangliosides were extracted in chloroform! methanol/water, followed by HPTLC analysis, as reported above.

Cell death assays NS2OY cells were plated as above and grown on glass coverslips in 0.5% fetal calf serum for 24 h either in the presence or in the absence of effectors. Medium was re-

.1. Neurochem., Vol. 71, No. 6, 1998

2316

R. MISASI ET AL.

FIG. 1. Scanning confocal microscopic analysis of prosaposin—ganglioside association on NS2OY cell surface. Cells were labeled either with rhodamine (TRITC)-conjugated CT B subunit for a direct staining or with anti-GM3 (GMR6) or with anti-GM2 (GMB28), followed by addition of goat anti-mouse 1gM conjugated with Texas Red for an indirect staining. After washing with PBS, cells were incubated with anti-769 Ab, followed by addition of FITC-conjugated goat anti-rabbit lgG. x750, bar = 2 ~.tm.a: Cells stained with anti-769 Ab, followed by addition of FITC-conjugated goat anti-rabbit lgG (a-i, a-2, and a-3). b: Cell stained with anti-GM3 mAb, followed by addition of goat anti-mouse 1gM conjugated with Texas Red (b-i); cell stained with anti-GM2 mAb followed by addition of goat anti-mouse gM conjugated with Texas Red (b-2); and cell stained with rhodamine (TRITC)-conjugated CT B subunit (b-3). C: Dual immunolabeling of anti-GM3 (red) and anti-769 Ab (green) (c-i), anti-GM2 (r) and anti-769 Ab (g) (c-2), and CT B subunit (red) and anti-769 Ab (green) (c-3). d: Two-dimensional scatter plot analysis of the dual-labeled fluorochromes (pseudocolor) GM3—prosaposin (d-i), GM2— prosaposin (d-2), and GM1—prosaposin (d-3). Diagrams show the pixel intensity distribution of a dual-channel section series. The xaxis represents intensity from the red channel; the y-axis represents intensity from the green channel. When the fluorochromes label the same area of the sample, the scatter plot has one cluster (d-1, yellow arrow); when there are no areas of colocalized fluorochromes, the scatter plot shows two distinct clusters (d-2 and -3, green and red arrows). e: Dual immunofluorescence image analysis represents exclusively the colocalization areas of GM3—prosaposin (e-i), GM2—prosaposin (e-2), and GM1—prosaposin (e-3).

moved, and 0.2% trypan blue in PBS was added to each well. Blue-stained dead cells were scored as a percentage of the total on an inverted microscope, counting 400 cells in four areas of each well.

nostained with anti-GM3 (Fig. lb-i) or anti-GM2 (Fig. ib-2) mAb showed a high fluorescence, whereas

Statistical analysis

of ganglioside molecules over the cell surface (Fig. Ib). To determine whether a possible association ex-

Data on induction of neurite outgrowth in NS2OY cells

were tested for statistical significance by ANOVA. Differences between treatment means were analyzed by a Student— Newman—Keuls comparison test.

RESULTS To verify whether prosaposin and ganglioside molecules interact within the plasma membrane of NS2OY cells, we analyzed their distribution on the cell surface. Dual fluorescence experiments were performed using anti-769 Ab, anti-GM3 mAb, anti-GM2 mAb, or CT, a well-known ganglioside-binding ligand (van Heyningen, 1974). Our results, obtained using scanning confocal microscopy, showed that anti-769 staining ap-

peared uneven and punctate over the cell plasma membrane and neurites (Fig. la-1—3). NS2OY cells immuJ. Neurochem., Vol. 71, No. 6, 1998

a less intense CT staining was observed (Fig. lb-3). Most of the cells showed an uneven signal distribution

isted between prosaposin and gangliosides, we superimposed the double immunostaining of anti-769 Ab and that of anti-GM3, anti-GM2, or CT. Anti-769 and anti-GM3 double staining revealed yellow areas, corresponding to nearly complete colocalization, indicating that prosaposin molecules were localized in membrane domains enriched with GM3 molecules (Fig. Ic-!). In contrast, very few colocalization areas were evident in anti-769—anti-GM2 (Fig. lc-2) and in anti-769— CT double-stained samples (Fig. lc-3). Scatter plot diagrams of the corresponding images showed graphically how the dual labels are localized. Figure Id-i, corresponding to Fig. lc-i, showed a single cluster of high pixel values (yellow arrow), indicating the presence of plasma membrane colocalization areas. Figure !d-2, corresponding to Fig. lc-2, and Fig. ld-

PROSAPOSIN-GM3 COMPLEXES IN NEURAL CELLS

2317

gether, these findings suggest that prosaposin and GM3 were nearly completely colocalized on the plasma membrane of NS2OY cells, whereas prosaposin and GM2 or GMi were rarely colocalized.

To verify whether GM3 binds directly to prosaposin, cell-free lysates from NS2OY neuroblastoma cells were immunoprecipitated with the R anti-saposin C Ab. Acidic glycosphingolipids were then extracted from

the immunoprecipitates, and HPTLC analysis showed that GM3 was the only ganglioside present (Fig. 3A,

lane b). To assess the specificity of GM3 binding to prosaposin, the antibody (R anti-saposin C Ab) was preabsorbed with prosaposin and then used for the immunoprecipitation. GM3 was not detected in the samples extracted from the immunoprecipitates obtained by such treatment. These findings suggested an associ-

FIG. 2. Scanning confocal microscopic analysis of prosaposin— ganglioside association on neurite-bearing NS2OY cells. a: Cell labeled with anti-GM3 mAb followed by addition of goat antimouse gM conjugated with Texas Red. b and d: Cells labeled with anti-769 Ab, followed by addition of FITC-conjugated goat anti-rabbit lgG. c: Cell labeled with rhodamine (TRITC)-conjugated CT B subunit. e: Dual immunolabeling of anti-GM3—anti769 Ab. f: Dual immunolabeling of CT—anti-769 Ab. a—d, x450, bar = 5 pm. e and f, 2