Ganglioside Modulation Regulates Epithelial Cell Adhesion and Spreading via Ganglioside-specific Effects on Signaling*

Gangliosides are implicated in regulating cell adhesion and migration on fibronectin by binding with the α5 subunit of α5β1integrin. However, the effects of gangliosides on cell spreading and related signaling pathways are unknown. Increases in gangliosides GT1b and GD3 inhibited spreading on fibronectin, concurrent with inhibition of Src and focal adhesion kinase. Although antibody blockade of GT1b or GD3 function and gene-modulated ganglioside depletion stimulated spreading and activated Src and focal adhesion kinase, the augmented spreading by disruption of GT1b function, but not by disruption of GD3 function, was inhibited by blockade of Src and focal adhesion kinase activation. In contrast, inhibitors of protein kinase C prevented the stimulation of spreading by GD3 functional inhibition, but not by GT1b functional blockade. Modulation of either GT1b or GD3 content affected phosphoinositol 3-kinase activation, and inhibition of this activation reversed the stimulation of cell spreading by anti-GD3 antibody, anti-GT1b antibody, and ganglioside depletion, suggesting that phosphoinositol 3-kinase is an intermediate in both the FAK/Src and protein kinase C pathways that lead to cell spreading. These studies demonstrate that epithelial cell ganglioside GT1b modulates cell spreading through α5β1/FAK and phosphoinositol 3-kinase signaling, whereas GD3-modulated spreading appears to involve phosphoinositol 3-kinase-dependent protein kinase C signaling.

Cell adhesion to the extracellular matrix triggers a cascade of intracellular biochemical signals regulated by the integrin family of receptors. Engagement and clustering of these integrins lead to the formation of focal adhesions, through which integrins are linked to intracellular cytoskeletal complexes, allowing cell spreading and migration (for review, see Ref. 1). One component of focal contact sites is focal adhesion kinase (FAK), 1 a nonreceptor tyrosine kinase that interacts with and is activated by the ␤ 1 subunit of integrin. Activation of FAK involves autophosphorylation at its Tyr 397 site, generating a high affinity binding site for the Src homology 2 domain. FAK is also known to interact directly with the p85 subunit of phosphoinositol 3-kinase (PI3K), enabling signal transduction. Activation of FAK, Src kinase, and PI3K has all been associated with stimulation of cell attachment, spreading, and migration (for review, see Ref. 1). Up-regulation of protein kinase C (PKC) also enhances cell adhesion, spreading, and migration on a fibronectin (FN) matrix by a mechanism that does not involve FAK phosphorylation (2)(3)(4).
Gangliosides are membrane glycosphingolipids that modulate several biologic processes of keratinocytes and keratinocyte-derived cell lines in vitro, including cell proliferation, adhesion, migration, differentiation, and apoptosis, at least in part by affecting transmembrane signaling (5)(6)(7)(8)(9)(10)(11). Keratinocytes 2 and the keratinocyte-derived SCC12 cell line share three major gangliosides, GM3, GD3, and GT1b (8), all components of the "b" pathway of ganglioside biosynthesis. This content of membrane gangliosides is dictated by the expression of enzymes that promote synthesis and metabolism, particularly glycosyltransferases and ganglioside-specific sialidase. The effect of gangliosides and ganglioside depletion on cell spreading and on signaling that may impact on spreading, particularly FAK and Src signaling, has received little attention. We have manipulated ganglioside expression exogenously by pharmacologic addition or anti-ganglioside antibody blockade of function, and endogenously by overexpression of genes that promote depletion of gangliosides or synthesis of the complex ganglioside GT1b. These studies show that depletion of specific gangliosides stimulates cell spreading, whereas increased membrane content of gangliosides inhibits it. The cell spreading inhibited by GT1b overexpression or induced by anti-GT1b antibody is mediated exclusively by the FAK/Src/PI3K pathway, whereas the stimulation of spreading induced by anti-GD3 antibody appears to involve the PKC pathway as well. demonstrated by Northern blot, sialidase expression by sialidase activity measurements, and ganglioside expression by thin layer chromatography (TLC) immunostaining as described (13), compared with parental SCC12 cells and the pcDNA3 mock transfectant control (pcDNA). Because of anticipated inhibition of SCC12 cell proliferation and decreased cell survival by overexpression of complex gangliosides, particularly GT1b, GM2/GD2 synthase cDNA (GenBank accession no. M83651, courtesy of Dr. Lloyd, New York) (14) was introduced using an inducible expression system (15,16) that utilizes anti-progestins as inducers. GM2/GD2 synthase cDNA in a pcDNA3 vector was treated with XhoI and ScaI restriction enzymes, then ligated into the p17x4-tkA (neomycin-resistant) vector opened at XhoI and EcoRV (Fig. 1, A  and top section of B). The ␤-galactosidase (␤-gal) cDNA in a pBluescript vector was linearized by treatment with PmeI and ApaI restriction enzymes and inserted into the same p17x4-tkA vector opened at EcoRV and ApaI sites (Fig. 1, A and bottom section of B) to generate a marker reporter construct. The pCEP4⅐GL-VP transactivator construct was a generous gift from Drs. S. Tsai and X. Wang, Houston, TX (Fig. 1C). The purified p17x4-tkA/GM2/GD2 synthase or p17x4-tkA/␤-gal construct was cotransfected into SCC12 cells with the purified pCEP4⅐GL-VP (hygromycin-resistant) transactivator construct using LipofectAMINE reagent. Putative cotransfectants were selected by antibiotic resistance to both 500 g/ml G418 and 60 g/ml hygromycin B, followed by limiting dilution to ensure clonality.
Ganglioside ELISAs-Parental SCC12 cells, mock transfectants (p17x4-tkA-pGL-VP cells), and p17x4-tkA/GM2/GD2 synthase-pGL-VP cells treated with or without 100 nM RU486 for 48 h were trypsinized, and suspended single cells were plated into 96-well microtiter plates. Cells were grown in the presence or absence of 100 nM RU486 for an additional 6 -8 h. Cells were then fixed with 4% formalin, followed by incubation with 10 -30 g/ml anti-ganglioside antibodies as used for TLC studies for 1 h and horseradish peroxidase-labeled anti-mouse IgM (Biodesign International, Saco, Maine) for 45 min. Ganglioside expression was detected by applying BM Blue POD substrate (Roche Molecular Biochemicals), and the absorbance was read at 450 nm in a UV max kinetic microplate reader (Molecular Devices, Menlo Park, CA). In each case, uncoated wells, omission of the anti-ganglioside antibody, and use of purified mouse IgM in place of anti-ganglioside antibody served as negative controls. Results of treatment with the blue substrate alone were subtracted from the readings for final data.
Cell Adhesion Assays-96 well plates were coated with or without 5 g/cm 2 cell-binding fragment of FN or with 10 g/cm 2 poly-L-lysine in phosphate-buffered saline (pH 7.4) overnight at 4°C. After removing unbound matrix solution, the plate was air dried at room temperature. After washing three times with phosphate-buffered saline, plates were incubated with 1% bovine serum albumin and phosphate-buffered sa-FIG. 1. GM2/GD2 synthase cDNA in an inducible p17x4-tkA vector system and the RU486 transactivator constructs. GM2/ GD2 synthase cDNA was cleaved from its pcDNA vector by treating with both XhoI and ScaI restriction enzymes and subcloned into the p17x4-tkA vector, opened at its XhoI and EcoRV sites (A and top row of B). Reporter gene ␤-gal cDNA was cleaved from its pBluescript vector by treating with both PmeI and ApaI restriction enzymes and subcloned into the p17x4-tkA between its EcoRV and ApaI sites (A and bottom row of B). The GM2/GD2 synthase cDNA-or ␤-gal cDNA-containing constructs were cotransfected with the transactivator pCEP4⅐GL-VP (C) into SCC12 cells. The function of the inducible system in SCC12 cells was tested preliminarily by cotransfection of the reporter gene ␤-gal cDNA in the p17x4-tkA vector (B, bottom row) and the transactivator pCEP4⅐GL-VP (C) into SCC12 cells. line for 1 h at 37°C to block unbound sites. SCC12 cells pretreated with or without ganglioside for 30 h, GM2/GD2 synthase transfectants pretreated for 30 h with 100 nM RU486 or vehicle, or SSIA cells were starved of fetal bovine serum and EGF to synchronize cycling for the final 18 h of incubation with ganglioside or RU486. In other studies, cells were starved for the 18 h and, during the final 30 min of starvation, were treated with anti-ganglioside antibodies to block ganglioside function. Cells were then transferred to the treated plates. Cell adherence was detected by staining with Rose Bengal dye, and the absorbance was read at 540 nm as described previously (6). All experiments were performed at least three times in triplicate.
Cell Spreading Assays-Eight-well plates were treated as described for adhesion assays above. SCC12 cells treated with or without gangliosides or anti-ganglioside antibodies and GM2/GD2 synthase transfectants and controls treated with or without RU486 as described for adhesion studies, or SSIA cells and controls (10 6 cells) were starved for 18 h in DMEM/F-12 medium without fetal bovine serum or EGF, then plated onto the treated plates. After incubation for 1.5 h at 37°C, cells were examined by phase contrast microscopy at ϫ200 magnification (model TMS-F, Nikon). Cells were classified as unspread (round shape with uneven outline), or spread (long and thin) based on cell morphology as described (17). At least 10 random fields were photographed and counted (about 80 -100 cells/field). The percentage of spread cells was determined as 100 ϫ spread cells/total cells (spread cells ϩ unspread cells) and expressed as the mean Ϯ S.D. of studies performed at least six times and assessed each time by three different individuals.
Immunoblotting-Immunoblotting was carried out as described (11) using an enhanced chemiluminescence (ECL) detection system (PerkinElmer Life Sciences) with whole cell lysates. Cells were lysed in boiling lysis buffer, and the insoluble material was removed by centrifugation. Total cell protein in the whole cell lysates was determined through a colorimetric assay (Bio-Rad) to ensure equal loading. Eight to 20 g of total protein from the whole cell lysates was boiled in Laemmli buffer (18) and loaded onto SDS-PAGE mini-gels. After transfer to polyvinylidene difluoride or nitrocellulose membranes, the separated proteins were detected by immunoblotting with anti-phosphotyrosine, -p85 of PI3K, -FAK (Transduction Labs, Lexington, KY), -Src, or -Srcphosphotyrosine 416 (Upstate Biotechnology, Lake Placid, NY) antibodies. Blots were reprobed as described previously (11) with anti-actin antibody to confirm equal loading. All blots were repeated at least three times with different preparations.
Immunoprecipitation-After starvation of serum, FN, and EGF overnight, cells were stimulated with 10 g/ml soluble cell-binding fragment of FN for 30 min. Cells were harvested and lysed in cold immunoprecipitation buffer (10). Total protein from the cell lysates (500 g-1 mg) was mixed with 5 g of anti-FAK, anti-Src, or anti-p85 of PI3K antibodies, and the total reaction volume was adjusted to 1 ml in the immunoprecipitation buffer. After incubation with the antibodies for 2 h at 4°C, protein A-agarose was added and incubated for an additional 30 min at 4°C (9). For immunoprecipitations with monoclonal antibody (anti-p85 of PI3K), 5 g of rabbit anti-mouse IgG antibody was applied to each sample for an additional 30 min before adding the protein A-agarose.
Blockade of FAK Function by Antisense Oligodeoxynucleotide-Synthetic oligodeoxynucleotides were generated complementary to either the sense or the antisense strand of the 20 nucleotides encoding human FAK at the 5Ј-end (19), including the initiator codon ATG (5Ј-ATG-GCAGCTGCTTACCTTGA-3Ј, FAK sense oligodeoxynucleotide; 5Ј-TCAAGGTAAGCAGCTGCCAT-3Ј, FAK antisense oligodeoxynucleotide). Cells were grown on eight-well Lab-Tek chamber slides for immunostaining, in six-well cell culture plates for immunoblotting, or in T-75 flasks for cell spreading studies as described above. Cells were incubated with DMEM/F-12 medium containing the oligodeoxynucleotides at 30 M at 37°C for 30 min, and then fetal bovine serum was added to a final concentration of 10%. The oligodeoxynucleotides were refreshed every other day for the 3-7 days before testing. FAK expression and activity were monitored and detected by immunofluorescence and immunoblotting with anti-FAK, anti-phosphotyrosine 20, and anti-phospho-FAK (Tyr 397 ) antibodies (Upstate Biotechnology). Cell spreading was determined as described above after blockade of FAK expression and function with FAK antisense oligodeoxynucleotides.
Modulation of FAK Expression and Activity by Gene Transfection-SCC12 cells were transiently transfected with either hemagglutinintagged wild FAK cDNA in a pcDNA vector or hemagglutinin-tagged FAK 397 site mutation cDNA (Y397K) in a pcDNA vector (both a generous gift from Dr. D. Schlaepfer, Scripps Research Institute, La Jolla, CA) using Lipofectin reagent (Invitrogen) according to the manufactur-er's instructions. Expression of the cDNAs was assessed by immunoblotting with anti-hemagglutinin and antibody (Upstate Biotechnology). To assess the function of FAK on cell spreading, untreated cells and cells transiently transfected with either wild type FAK or Y397K mutated FAK cDNA were harvested after treatment with either gangliosides or anti-ganglioside antibodies as indicated above, and spreading assays were performed.
FAK Expression, Activity, and Phosphorylation-Immunoprecipitated FAK was incubated with 50 l of buffer containing 50 mM HEPES (pH 7.4), 10 mM MnCl 2 , and 50 g of polyglutamic acid:polytyrosine (4:1) for 10 min at 25°C, then 5 M ATP and 5 Ci of [␥-32 P]ATP (3,000 Ci/mmol) were added to initiate the reaction. Incubation was terminated 10 min later by adding 50 l of 20% trichloroacetic acid, and the unincorporated radiolabeled ATP was removed by applying the precipitate onto microcentrifuge filter paper. After washing three times with 10% trichloroacetic acid, the filter paper loaded with the product was air dried, placed in scintillation fluid, and ␥-32 P-labeled substratepolyglutamic acid:polytyrosine (4:1) was measured in a Beckman LS 6000 scintillation counter. FAK activity assays were performed three times in triplicate.
To determine FAK expression and phosphorylation, SCC12 cells were pretreated with either gangliosides or anti-ganglioside antibodies, then stimulated with 10 g/ml cell-binding fragment of FN (Upstate Biotechnology) for 30 min at 37°C. FAK was immunoprecipitated from whole cell lysates using rabbit anti-human FAK polyclonal antibody, and the denatured precipitates were applied to a 10% SDS-PAGE mini-gel. FAK phosphorylation was detected with mouse anti-phosphotyrosine (PY-20) antibody and the phosphorylation of FAK specifically at its Tyr 397 site was detected with rabbit anti-phospho-FAK (Tyr 397 ) antibody. To confirm equal loading of immunoprecipitated FAK, the membrane was reprobed with mouse anti-human FAK monoclonal antibody. FAK expression was detected by applying 10 g of total protein from whole cell lysate to a 10% SDS-PAGE mini-gel and immunoblotting with anti-FAK monoclonal antibody.
Src Kinase Expression, Activity, and Phosphorylation-Endogenous c-Src was immunoprecipitated from cells pretreated with or without gangliosides or anti-ganglioside antibodies. The immunoprecipitated c-Src kinase was resuspended in 15 l of Src kinase buffer (20 mM PIPES (pH 7.0), 10 mM MgCl 2 , 1 mM dithiothreitol) containing 2.5 g of acid-denatured enolase (Sigma) as described (20). Kinase reactions were initiated by the addition of 10 M ATP and 10 Ci of [␥-32 P]ATP (3,000 Ci/mmol) for 15 min at 30°C. Reactions were stopped by boiling in Laemmli buffer (18) for 10 min, and proteins were separated on a 10% SDS-PAGE mini-gel. The enolase band was visualized by autoradiography on Kodak X-Omat film developed at Ϫ70°C overnight and quantified by a Storm 800 Fluorescence PhosphorImager (Molecular Dynamics, Inc.). Src kinase activity was also determined by measuring the phosphorylation of a Src substrate peptide (KVEKIGEGTYGVVYK) using immunoprecipitated c-Src with a Src kinase assay kit (Upstate Biotechnology) according to the manufacturer's instructions. The experiment was performed three times in triplicate.
To determine the expression and phosphorylation of c-Src kinase, cells were pretreated with either gangliosides or anti-ganglioside antibodies, washed with DMEM/F-12 medium, and stimulated with 10 g/ml cell-binding fragment of FN for 30 min at 37°C. Whole cell lysates (25 g of protein) were applied to a 10% SDS-PAGE mini-gel and then immunoblotted with anti-Src-phosphotyrosine 416 polyclonal antibody (Upstate Biotechnology) to assess c-Src phosphorylation. Gels were reprobed with anti-Src monoclonal antibody to measure c-Src expression as described before (11).
PI3K Expression and Activity-To determine PI3K expression and activity with FN stimulation, SCC12 cells were pretreated with either gangliosides or anti-ganglioside antibodies, washed, and stimulated with FN as described for the assays of FAK. PI3K activity was assayed as described previously (21,22). Briefly, immunoprecipitated PI3K was incubated with 50 l of kinase buffer containing 0.2 mg/ml phosphatidylinositol, then treated with 20 Ci of [␥-32 P]ATP (3,000 Ci/mmol) and 20 mM MgCl 2 for 10 min. The reaction was terminated by adding 150 l of chloroform, methanol, 11.6 M HCl (v/v/v, 50:100: 1), and phosphatidylinositol was extracted with chloroform. The organic phase was washed with methanol and 1 M HCl (v/v, 1:1) and lyophilized. Phosphatidylinositol was resolved by TLC in chloroform, methanol, 28% ammonium hydroxide, and water (v/v/v/v, 86:76: 10: 14) for 45 min at 20°C. Phosphorylated products were visualized by autoradiography with Kodak X-Omat film developed at Ϫ70°C overnight and quantified by a Storm 800 Fluorescence PhosphorImager. PI3K expression was determined by separation of 15 g of total protein from whole cell lysate proteins on a 10% SDS-PAGE mini-gel and immunoblotting with anti-PI3K p85 antibody. Experiments were performed four times.
Blockade of Specific Signaling Pathways with Inhibitors-After pre-treatment with or without ganglioside or anti-ganglioside antibody in the presence or absence of inhibitors as indicated below, cells were plated onto FN or poly-L-lysine-coated plates as described above for spreading assays. During the 1.5-h incubation on the plates, cells were treated continuously with or without 20 M LY294002 (PI3K inhibitor) (Calbiochem), 3 M {{4-amino-1-tert-butyl-3-(1Ј-naphthyl)pyrazolo [3,4d]pyrimidine}} (PP1, Src inhibitor) (Alexis, San Diego), or 20 g/ml anti-␤ 1 integrin blocking antibody (Chemicon International, Temecula, CA), and spreading was measured as described above. To explore the effect on ganglioside-modulated spreading of PKC inhibition, SCC12 cells were treated with or without PKC inhibitors, 100 nM bisindolylmaleimide I or 1 M calphostin C (Calbiochem) for 3 h in DMEM/F-12 medium containing 6 mg/ml bovine serum albumin. During the final 30 min of incubation, cells were treated with 10 g/ml anti-ganglioside antibodies, including anti-GD3, anti-GT1b and, as a control, anti-GM2 antibody before cell replating and assessment of spreading. 100 g/ml purified mouse IgM also served as a negative control.

RESULTS
Generation of Stable Inducible GM2/GD2 Synthase Transfectants-Preliminary studies with RU486-inducible expression of ␤-gal showed strong expression of the ␤-gal, peaking at 24 -48 h after exposure to 100 nM RU486, totally reversible by elimination of the RU486 within 72 h. 29 monoclonal cell lines of GM2/GD2 synthase-expressing cells were generated. Northern analyses showed a 3.6 -5.9-fold increase in expression of GM2/GD2 synthase in cells transfected with both the transactivator and the GM2/GD2 synthase constructs when induced by RU486 ( Fig. 2A). Expression of the GM2/GD2 synthase in transfected cells without RU486 induction and in parental SCC12 cells and mock transfected cells with and without RU486 induction were equivalent. TLC immunostaining (not shown) and ganglioside ELISA (Fig. 2B) showed increased expression of GT1b and de novo expression of GD2 in transfected cells induced by RU486, whereas expression of gangliosides GM3 and GD3 was unchanged. Transfected cells did not express GM2 or GM1 de novo. Control cells (parental SCC cells, mock transfected cells with and without induction, transfected cells without the RU486 induction, and cells treated with vehicle only) showed no change in ganglioside expression. Four SSIA cell lines (SSIA3, 6, 12, and 25), four GM2/GD2 synthase lines (clones 1, 5, 11, and 26) with and without RU486 induction, and two of each mock transfected cell lines were studied. Results were consistent among transfected cell lines in each experiment.
Gangliosides Modulate Cell Adhesion on FN Matrix-Consistent with our previous studies that showed GT1b to inhibit adhesion of normal undifferentiated cultured keratinocytes to FN (6), addition of 1 M GT1b significantly inhibited attachment of the SCC12 cells to FN (p Ͻ 0.001) (Fig. 3A). 1 M GD3 also inhibited adhesion to FN, although slightly less than the inhibition by GT1b. Adhesion to FN was not impacted by treatment with 50 M GM3 or GM2. Gangliosides had no effect on SCC12 cell adhesion to poly-L-lysine or plastic (not shown). Blockade of functional ganglioside resulted in increased adherence to FN with anti-GT1b (p Ͻ 0.05), anti-GD3 (p Ͻ 0.05), and the combination of anti-GT1b and anti-GD3 (p Ͻ 0.01) antibodies (Fig. 3B). No significant change in adhesion was induced by antibodies directed against other gangliosides of the SCC12 cell (anti-GM3 and anti-9-O-acetyl-GD3 antibodies), or when cells were plated on poly-L-lysine. Endogenous ganglioside modification by RU486-induced overexpression of GM2/GD2 synthase (increases GT1b without significantly decreasing GD3 or GM3) decreased adhesion to FN in contrast to parental SCC12 cells, vector controls, and transfected cells without RU486 stimulation (p Ͻ 0.01) (Fig. 3C); SSIA cells (overexpression of ganglioside-specific sialidase and depletion of all gangliosides) showed significantly increased adherence to FN (p Ͻ 0.001) compared with parental SCC12 cells and pcDNA vector controls (Fig. 3C). Neither induced GM2/GD2 synthase overexpressors nor SSIA cells showed any change in adherence to a poly-L-lysine matrix. Consistent with the time-limited effect of adding GT1b to keratinocytes plated on FN (6), the significant increase in binding of SSIA cells to FN was an early phenomenon, with a maximal difference between SSIA cells and controls at 90 min but no difference between binding of SSIA cells and controls after 120 min (Fig. 3D).
Gangliosides Modulate Cell Spreading on FN Matrix-To assess the effect of keratinocyte plasma membrane gangliosides on cell spreading, keratinocyte-derived SCC12 cells were pretreated with or without gangliosides or anti-ganglioside antibodies. In addition, ganglioside-deficient SSIA cells, their mock pcDNA cell controls, GM2/GD2 synthase overexpressors with or without RU486 induction, or their vector controls were allowed to attach and spread on FN matrix in the presence of soluble cell-binding fragment of FN for 1.5 h. Ganglioside depletion by sialidase overexpression (Fig. 4, B and E) or blockade of specific keratinocyte ganglioside function with anti-GT1b (Fig. 5, D and G) or -GD3 antibody (Fig. 5, F and G) increased cell spreading on a FN matrix by 2.5-fold, 2.2-fold, and 2.0-fold, respectively, compared with parental SCC12 cells (Figs. 4, A and E; and 5G), pcDNA control cells (Fig. 4E), or cells treated with control anti-ganglioside antibodies (Fig. 5, B and G). In contrast, pharmacological addition of ganglioside GT1b (Fig. 5, C and H) or GD3 (Fig. 5, E and H) or overexpression of GM2/ GD2 synthase (Fig. 4, D and E) inhibited cell spreading on a FN matrix by 2.9-fold, 2.4-fold, and 2.7-fold, respectively, compared with parental SCC12 cells, GM2/GD2 synthase transfected cells without RU486 induction (Fig. 4, C and E), or control gangliosides (Fig. 5, A and H). None of the ganglioside modulations affected cell spreading on either poly-L-lysine (Figs. 4E and 5, G and H) or the uncoated plastic surface (not shown).
Inhibition of Spreading by Specific Gangliosides Is Pathwaydependent-To explore the molecular mechanisms of the effect of gangliosides on keratinocyte spreading, the impact of modulation of integrin ␤ 1 subunit, c-Src, FAK, and PI3K function were examined. The effect of FAK function on inhibition of cell spreading by gangliosides was evaluated by transient transfection of wild type FAK cDNA into the SCC12 cells; the effect of FAK function on stimulation of spreading by specific or global ganglioside depletion was assessed by transient transfection of mutant FAK cDNA (Y397F) or incubation with FAK antisense oligodeoxynucleotide. The transfection efficiency of the wild type FAK cDNA was 49.2%, and that of the Y397F FAK cDNA was 42.7%. As shown in Fig. 6A, blockade of FAK function by either overexpression of FAK mutated at the critical Y397F phosphorylation site or incubation of cells with FAK antisense oligodeoxynucleotide decreased cell spreading on a FN matrix by 3.1-fold and 4.1-fold, respectively (Fig. 6A) compared with parental SCC12 cells. Conversely, overexpression of wild type FAK increased cell spreading by 2.1-fold. The stimulatory effect of wild type FAK overexpression on cell spreading was maintained in the presence of GT1b, but it could not reverse the inhibition of spreading induced by GD3 (Fig. 6B). Treatment of SCC12 cells with FAK antisense oligodeoxynucleotides or by transient transfection with Y397F mutant FAK cDNA markedly suppressed cell spreading (Fig. 6A)  sialidase overexpression or specific GT1b depletion by anti-GT1b antibody treatment (Fig. 6C). Blockade of FAK function was unable to reverse the stimulation of cell spreading triggered by treatment with anti-GD3 antibody (Fig. 6C).
Gangliosides Inhibit Phosphorylation and Activity of FAK, Src, and PI3K, but Not Their Expression-Blockade of specific ganglioside function by anti-GT1b or anti-GD3 antibody increased the phosphorylation of FAK at tyrosine 397 site by 2.6-fold and 2.5-fold, respectively (Fig. 8A, bottom row), and increased FAK activity by 1.9-fold and 1.7-fold, respectively (Fig. 8B). FAK phosphorylation at the tyrosine 397 site of SSIA cells was increased 3.4-fold, and activity was increased 2.3-fold compared with parental SCC12 and pcDNA controls (Fig. 8, A,  bottom row, and B). Consistently, pharmacological addition of ganglioside GT1b or GD3 decreased FAK phosphorylation at tyrosine 397 site and activity (Fig. 8A, bottom row, and B). Ganglioside GM2 treatment or functional blockade with anti-9-O-acetyl-GD3 antibody did not affect either FAK phosphorylation or activity (Fig. 8). FAK expression was not altered by ganglioside supplementation or functional blockade (Fig. 8A,  top row). The effect of gangliosides on FAK phosphorylation specifically at the 397 site was similar to that on phosphoryl-  Blockade of specific ganglioside function using anti-GT1b or anti-GD3 antibody or of total ganglioside function by sialidase overexpression increased the phosphorylation of c-Src at tyrosine 416 by 2.1-3.6-fold (Fig. 9A, bottom row) when cells were grown in the presence of FN; pharmacological addition of ganglioside GT1b or GD3 decreased Src kinase phosphorylation at tyrosine 416 by 2.1-fold or 1.5-fold, respectively (Fig. 9A, bottom  row). Treatment with anti-9-O-acetyl-GD3 control anti-ganglioside antibody or GM2 control ganglioside, and transfection with the pcDNA vector, had no effect on Src phosphorylation (Fig. 9A, bottom row). Ganglioside modulation did not alter Src kinase expression (Fig. 9A, top row). As measured by in vitro phosphorylation of acid-denatured enolase, Src kinase activity was increased dramatically in both ganglioside-depleted SSIA cells by 4.5-fold (Fig. 9B, lane 3) and by functional blockade of ganglioside GT1b or GD3 by 3.1-fold and 1.8-fold, respectively (Fig. 9B, lanes 4 and 5), compared with parental SCC12 cells (Fig. 9B, lane 1), mock pcDNA transfectant (Fig. 9B, lane 2), and cells with functional blockade of ganglioside 9-O-acetyl-GD3 (Fig. 9B, lane 6). Pharmacological addition of ganglioside GT1b or GD3 decreased Src kinase activity by 3.6-fold and 2.1-fold, respectively (Fig. 9B, lanes 8 and 9), whereas GM2 treatment had no effect on Src kinase activity (Fig. 9B, lane 7). Similar results were noted by measuring the phosphorylation of Src substrate peptide using a Src kinase assay kit (Fig. 9C).
Blockade of GT1b or GD3 function increased the activity of PI3K by 3.8-fold and 2.9-fold, respectively (Fig. 10A, top row), whereas sialidase overexpression increased PI3K activity by 5.1-fold. Pharmacological addition of ganglioside GT1b or GD3 diminished PI3K activity by 3.3-fold and 2.9-fold, respectively (Fig. 10B, top row). The inhibition of PI3K activity by gangliosides was comparable with the inhibition by treatment with anti-␤ 1 integrin antibody (2.1-fold) but was less than that induced by specific inhibition of PI3K activity with LY294002 (8.2-fold) (data not shown). Neither functional inhibition of 9-O-acetyl-GD3 nor addition of GM2 modulated PI3K activity (Fig. 10). PI3K expression was not affected by the addition or blockade of any ganglioside (Fig. 10, A and bottom rows of B).

Blockade of PKC Signaling Prevents the Stimulation of Spreading by Depletion of GD3 Function but Not of GT1b
Function-Treatment of SCC12 cells with the PKC inhibitors calphostin C (Fig. 11) or bisindolylmaleimide I (not shown) significantly reversed the effect of stimulation of cell spreading by anti-GD3 antibody, but it had no effect on the stimulation of cell spreading induced by anti-GT1b antibody in the presence of FN. DISCUSSION Our previous studies have demonstrated an inhibitory role of the more highly sialylated keratinocyte gangliosides, GT1b and GD3, in cell adhesion and migration specifically on a FN matrix (6,23). The mechanism of this inhibition, at least in part, relates to the specific ability of GT1b and to a lesser extent GD3, but not other tested gangliosides, to bind directly to the ␣ 5 subunit of ␣ 5 ␤ 1 , the primary receptor by which keratinocytes and keratinocyte-derived cell lines bind to FN, and the resultant suppression of the ␣ 5 ␤ 1 -FN interaction (9). Through this interaction with integrin, GT1b is also able to trigger cell apoptosis when keratinocytes and keratinocyte-derived SCC12 cells are plated on FN (23, 33) by a mechanism that involves inhibition of the integrin-linked kinase/protein kinase B/Akt signaling pathway (11,13). GD3, although also able to induce keratinocyte and keratinocyte-derived cell apoptosis through a mechanism that promotes mitochondrial cytochrome c release (11), does not require cell exposure to FN and has no effect on integrin-linked kinase/protein kinase B signaling (11). These studies of apoptosis provide evidence of two independent mechanisms by which GT1b and GD3 influence signaling in keratinocytes.
Here, we show that both GT1b and GD3, but not other keratinocytes or keratinocyte-derived SCC12 cell gangliosides, are able to inhibit cell spreading on a FN matrix, and we have delineated the signaling mechanisms responsible for this inhibition. The increase in cell spreading observed when cells were treated with anti-GT1b and anti-GD3 antibodies, but not with antibodies directed against other gangliosides, further confirms the specificity of the effect on cell spreading.
Although pharmacologic addition of gangliosides increases membrane concentration of added gangliosides, and anti-ganglioside antibodies block ganglioside function, physiologically relevant shifts in membrane ganglioside content likely result from changes in expression of enzymes that deplete gangliosides or lead to the synthesis of more complex gangliosides. As a result, we also examined the effects on cell spreading of endogenous modification in gangliosides through gene overexpression of two human genes: a ganglioside-specific plasma membrane sialidase, which we have previously shown to deplete totally the membrane gangliosides of SCC12 cells (13), and GM2/GD2 synthase, which we predicted would lead to endogenous increases in GT1b content in the SCC12 cells. Consistent with the effects on spreading of exogenous manipulation, depletion of membrane gangliosides increased spreading and the increased GT1b expression after stable GM2/GD2 synthase transfection reduced cell spreading when cells were plated on FN in the absence of both serum and growth factors.
In previous studies, we noted no effect of GM3, the predominant ganglioside of SCC12 cells, on normal keratinocyte or , and SCC12 cells pretreated with or without either 10 g/ml anti-GT1b, -GD3, or as a control, anti-9-Oacetyl-GD3 antibody, or with or without 1 M GT1b, 1 M GD3, or as a control, 50 M GM2 were stimulated with 10 g/ml cell-binding fragment of FN and lysed as described under "Experimental Procedures." c-Src expression was detected using 25 g of total protein from whole cell lysate (A, top row), and c-Src phosphorylation was detected with anti-Src-phosphotyrosine 416 antibody (A, bottom row). c-Src kinase activity was measured by in vitro kinase assays with c-Src immunoprecipitates from cells treated as above using acid-denatured enolase as a substrate. The 32 P-labeled enolase band was visualized by autoradiography on Kodak X-Omat film (B). c-Src kinase activity was also assessed by measuring the phosphorylation of a substrate peptide and expressed as pmol of phosphate incorporated into the substrate peptide/ min/100 g of total cellular protein as indicated under "Experimental Procedures" (C). *, p Ͻ 0.05; **, p Ͻ 0.01; ***, p Ͻ 0.001. SCC12 cell adhesion to FN (6), in contrast to evidence that adhesion to FN of mouse mammary carcinoma cells requires GM3 (24). Here we provide further evidence of a lack of GM3 effect on spreading, FAK phosphorylation, and Src activation in keratinocyte-derived cells. Although the difference in effects of GM3 on adhesion may reflect cell specificity, the previously described effects of GM3 on adhesion may also be mediated via EGFR signaling and cross-talk with the FAK signaling pathway 2 (25,26); our studies were purposely performed in the absence of both growth factors and serum to address specifically the role of FAK/Src/PI3K signaling without the influence of autophosphorylation of EGFR, providing further evidence that any influence of GM3 on cell motility and spreading may result from EGFR signaling pathway.
Previous studies have demonstrated that activation of FAK and Src kinase signaling is important for epithelial cell spreading. We confirmed that the inhibition of cell spreading by GT1b involves the inhibition of activation of FAK and Src kinase. The inhibition of spreading by GD3, however, occurs by an additional pathway that includes PKC, so that GD3 is able to modulate spreading despite blockade of FAK or Src activation. The role of gangliosides in Src signaling has also received scant attention. Recently, Li et al. (27) demonstrated that ganglioside GD1a induces phosphorylation of Src kinase in human dermal fibroblasts in the absence of serum or growth factors. In con-trast, we have shown inhibition of Src kinase phosphorylation by a ganglioside in the SCC12 keratinocyte-derived cell line under these conditions in the presence of FN, suggesting both ganglioside and cell specificity of the effect on Src activation. PI3K-dependent, FAK-independent mechanisms for epithelial cell adhesion have been described previously (28), as has the role of activation of PKC for cell attachment and spreading (29). Although PI3K and PKC pathways appear to be distinct in several studies, recent investigations suggest a relationship of early PI3K activity as a requisite for the activation of certain isoforms of PKC (1), such as PKC-⑀ (30) and PKC- (31). Our preliminary data suggesting that GD3 is able to inhibit PKC activation via a PI3K-dependent cascade support a common pathway in epithelial cell spreading. The isozymes of PKC which are involved in this GD3-mediated effect on cell spreading and the time course of this activation deserve further investigation.
How do gangliosides GT1b and GD3 inhibit nonreceptor FAK and Src kinase signaling? Although gangliosides have a lipid component in the membrane that putatively could interact with FAK (cytoplasmic) and Src-kinase (located on the inner leaflet of the plasma membrane), the discrepant influences on spreading and FAK or Src signaling of specific gangliosides that differ only in their sialic acid and carbohydrate chain moieties suggest a direct ganglioside interaction external to the membrane with a receptor upstream of these signaling molecules. Furthermore, the direct relationships demonstrated between both GT1b and ␣ 5 ␤ 1 (9) and GM3 and the EGFR (10) require intact carbohydrate groups on both proteins and the gangliosides, providing further evidence that the external component of gangliosides is required, rather than the lipid intramembrane component. The recent recognition that signaling occurs in specific membrane domains that are enriched in cholesterol and glycosphingolipids provides a potential explanation for the separate signaling pathways inhibited by specific gangliosides and induced by their depletion. We propose that gangliosides GT1b and GD3 reside in distinct membrane microdomains with their specific signaling molecules, consistent with the recent demonstration by Vyas et al. (32) that GM1 and GD3 are "packaged" in different membrane regions. Thus, GT1b may be packaged with ␣ 5 ␤ 1 integrin, FAK, Src kinase, FIG. 10. Ganglioside depletion or functional blockade of ganglioside GT1b or GD3 activates PI3K activity, whereas exogenously increased GT1b or GD3 inhibits PI3K activity. Ganglioside-depleted SSIA cells (four cell lines), mock transfected pcDNA cells (two cell lines), and SCC12 cells pretreated with or without either gangliosides or anti-ganglioside antibodies as described above were starved for 18 h in serum-free DMEM/F-12 medium. Cells were lysed after stimulation with 10 g/ml soluble cell-binding fragment of FN for 30 min. PI3K activity was measured using phosphatidylinositol as a substrate with immunoprecipitated PI3K as described under "Experimental Procedures," shown in the top rows in both A and B. Expression of PI3K was determined using 15 g of total protein from whole cell lysate and as shown in the bottom rows of both A and B. The SSIA cell and pcDNA phosphorylations shown in A are representative of the studied cell lines.
FIG. 11. Inhibition of PKC signaling reverses the stimulatory effect of anti-GD3 antibody on SCC12 cell spreading on a FN matrix. SCC12 cells were incubated with either 100 nM bisindolylmaleimide I (not shown) or 1 M calphostin C for 3 h, and anti-ganglioside antibodies were added in the final 30 min. Cells were then harvested and plated onto eight-well cell culture plates precoated with 5 g/cm 2 soluble cell-binding fragment FN. Cell spreading was measured as described under "Experimental Procedures," and results are expressed as the mean Ϯ S.D. *, p Ͻ 0.05. and PI3K, whereas GD3 is complexed within membranes with PKC and PI3K, allowing differential inhibition of signaling pathways, leading to apoptosis and inhibition of spreading by different mechanisms.