Exogenous ganglioside GD1a enhances epidermal growth factor receptor binding and dimerization.

Gangliosides are shed by tumor cells and can bind to normal cells in the tumor microenvironment and affect their function. Exposure of fibroblasts to exogenous gangliosides increases epidermal growth factor (EGF)-induced fibroblast proliferation and enhances EGF receptor (EGFR)-mediated activation of the mitogen-activated protein kinase signaling pathway (Li, R., Liu, Y., and Ladisch, S. (2001) J. Biol. Chem. 276, 42782-42792). Here we report that the EGFR itself is the target of this ganglioside effect: Preincubation of normal human dermal fibroblasts with G(D1a) ganglioside enhanced both EGF-induced EGFR autophosphorylation and receptor-tyrosine kinase activity. The enhancement was rapid (within 30 min), not due to alteration of time kinetics of the EGFR response to EGF, and reproduced in purified G(D1a)-enriched cell membranes isolated from ganglioside-preincubated fibroblasts. Evaluating the initial steps underlying activation, EGF binding, and EGFR dimerization, we found that G(D1a) enrichment of the cell membrane increased EGFR dimerization and the effective number of high affinity EGFR without increasing total receptor protein. Unexpectedly, G(D1a) enrichment also triggered increased EGFR dimerization in the absence of growth factor. This resulted in enhanced activation of the EGFR signal transduction cascade when EGF was added. We conclude that membrane ganglioside enrichment of normal fibroblasts (such as by tumor cell ganglioside shedding) facilitates receptor-receptor interactions (possibly by altering membrane topology), causing ligand-independent EGFR dimerization and, in turn, enhanced EGF signaling.

Cell Culture-NHDF purchased from Clonetics (San Diego, CA) were cultured in fibroblast complete growth medium, FGM-2 (Clonetics), which contains 2% fetal bovine serum and 0.5 ml each of insulin, human fibroblast growth factor, and GA1000 per 500 ml. The culture medium was changed every 3 days. All experiments were performed using subconfluent cultures of passages 3-10. Cell viability was assessed by trypan blue dye exclusion. For serum-free culture, fibroblast basal medium (FBM) was used.
Preparation of Cell Lysate-1-2 ϫ 10 5 NHDF were seeded per 100-mm dish or per well in 6-well plates in FGM-2. Upon reaching subconfluence, the cells were incubated with gangliosides in FBM for 6 or 18 h. Then the culture medium was removed, and the cells were washed twice with FBM to remove unbound ganglioside and exposed to 2 ng/ml EGF in FBM for 5 min at 37°C (29). Then the cells were washed twice with ice-cold phosphate-buffered saline (PBS) and lysed for 20 min in lysis buffer, 1 ml/100-mm dish or 300 l/6-well culture plate. The lysis buffer contained 20 mM Tris, pH 7.5, 150 mM NaCl, 1 mM EDTA, 1 mM EGTA, 1% Triton X-100, 2.5 mM sodium pyrophosphate, 1 mM ␤-glycerol phosphate, 1 mM Na 3 VO 4 , 1 g/ml leupeptin, and 1 mM phenylmethylsulfonyl fluoride. The lysate was transferred to microcentrifuge tubes, sonicated briefly on ice, and centrifuged at 10,000 ϫ g for 10 min at 4°C. The supernatant was adjusted to 1 g of protein/l and used for the kinase assays. Proteins were quantified by the Lowry method using bovine albumin as a standard (30).
Cell Preparation for the EGFR Inhibitor Assay-To test for the specificity of the effect of ganglioside preincubation on receptor autophosphorylation, the EGFR inhibitor, AG1478, was used. NHDF were exposed to G D1a in FBM for 18 h. During the last 3 h, 1 M AG1478 or Me 2 SO was added (31). Then the cells were exposed to 2 ng/ml EGF for 5 min and harvested, and the cell lysate was prepared as above.
EGFR Autophosphorylation Assay-To preclear nonspecific binding, 200 l of cell lysate (ϳ200 g of total protein) were mixed with 100 l of washed protein G-Sepharose-agarose bead slurry (50-l packed beads), stirred for 2 h at 4°C., and microcentrifuged at 10,000 ϫ g for 5 s. The supernatant was transferred to a new microcentrifuge tube, mixed with 4 g of sheep polyclonal IgG anti-EGFR antibody, and incubated overnight with gentle stirring at 4°C. The immune complexes were recovered by adding 100 l of the protein G-Sepharoseagarose bead slurry, gently rocking the mixture for 2 h at 4°C, and microcentrifuging at 14,000 ϫ g for 5 s. The supernatant was removed, the beads were washed 3 times with ice-cold lysis buffer, resuspended in 50 l of 2ϫ SDS sample buffer, boiled for 5 min, and microcentrifuged. 20 l of each supernatant (ϳ40 g) protein was loaded onto a 7.5% SDS-polyacrylamide gel. EGFR autophosphorylation was detected by Western blot using an anti-phosphotyrosine antibody Tyr(P) (PY99) (19), and total EGFR was detected by an anti-EGFR antibody.
Western Blotting-The sample was transferred to a polyvinylidene fluoride membrane, incubated in 25 ml of blocking buffer for 1 h at room temperature, and then incubated with primary antibody (rabbit polyclonal IgG; 1:1000 dilution) with gentle agitation overnight at 4°C. After 3 washes, the membrane was incubated in the medium containing horseradish peroxidase-conjugated anti-rabbit antibody (1:2000). Proteins were detected by chemiluminescence and compared with standard proteins of different molecular weights.
EGFR-tyrosine Kinase Activity-EGFR-tyrosine kinase activity was measured as the phosphorylation of an EGFR substrate peptide using a tyrosine kinase assay kit (Calbiochem). Briefly, the EGFR were immunoprecipitated from the cell lysate (ϳ200 g of protein). Half of the immunoprecipitated protein (10 l) was transferred to a microcentrifuge tube that contained 20 l of protein-tyrosine kinase reaction mix containing 30 M ATP, 50 M substrate peptide, 1 Ci of [ 32 P]ATP, and 4 l of 10ϫ protein-tyrosine kinase reaction buffer consisting of 200 mM magnesium chloride, 10 mM manganese chloride, 2 mM EGTA, 80 mM ␤-glycerophosphate, and 80 mM imidazole hydrochloride, pH 7.3. The mixture was incubated at 30°C for 10 min with agitation. The reaction was stopped by placing the tubes on ice followed by centrifugation at 10,000 ϫ g at 4°C for 5 s. The supernatant was transferred to a new tube containing 10 l of stop solution mix (8 M guanidine hydrochloride) and briefly centrifuged. 8 l of avidin solution was added to the supernatants, and the samples were incubated for 5 min at room temperature. 50 l of wash solution and 20 l of the reaction samples were transferred into the reservoirs of centrifugal ultrafiltration units, centrifuged for 5 min at 14,000 ϫ g, and washed 3 times with 100 l of wash solution. The washed filters were transferred to scintillation vials, scintillation mixture was added, and radioactivity was quantified. Net cpm of 32 P incorporated into the substrate peptide was calculated by subtracting the nonspecific binding of [ 32 P]ATP from the total cpm.
Phospho-p44/42 MAP Kinase Assays-Phosphorylation of p44/42 MAP kinase was determined by Western blot using the phospho-p44/42 MAP kinase antibody. A MAP kinase antibody was used to detect the total MAP kinase in each sample. Equal volumes of lysate and sample buffer were mixed, boiled for 5 min, vortexed, microcentrifuged for 2 min, and then subjected to SDS-PAGE electrophoresis (12% gel) and Western blotting.
Plasma Membrane Separation and in Vitro Assay of EGFR Autophosphorylation-NHDF were incubated with G D1a in FBM for 18 h. The plasma membranes were separated as previously described (32). Briefly, after aspirating the medium, the cells were washed twice with the ice-cold buffer A (0.25 M sucrose, 1 mM EDTA, 20 mM Tricine, pH 7.8), scraped into 3 ml of buffer A, centrifuged at 1000 ϫ g for 5 min, resuspended in 1 ml of buffer A, and homogenized in a 2-ml Wheaton tissue grinder with 20 strokes. The cell homogenates were centrifuged at 1000 ϫ g for 10 min. The pellets were resuspended in 1 ml of buffer A, homogenized, and centrifuged at 1000 ϫ g for 10 min. The two supernatants were combined, layered on 30% Percoll in buffer A, and ultracentrifuged at 84,000 ϫ g for 30 min, and the visible band was collected. These plasma membrane fractions were washed 3 times with HNG buffer (20 mM HEPES, 150 mM NaCl, 10% glycerol, pH 7.5), and the protein content was quantified. The resuspended pellets were assayed for EGFR activity (next paragraph).
In Vitro EGFR Activity Assays-10 g of protein of the plasma membrane preparation (15 l) was mixed on ice with 15 l of autophosphorylation buffer (HNG buffer containing 40 M ATP, 30 mM MnCl 2 , and 400 mM Na 3 VO 4 ). EGF was added to a final concentration of 2 ng/ml. Then the mixture was incubated at 37°C for 10 min with gentle agitation. The reaction was stopped by adding 30 l of 2ϫ lysis buffer, and the sample was stored on ice for 30 min. The samples were immunoprecipitated with the EGFR antibody (sheep polyclonal IgG), and the phospho-EGFR, total EGFR, and Tyr(P) kinase activities were tested as before.
Plasma Membrane Binding of [ 14 C]G D1a -To assess [ 14 C]G D1a binding, cells were cultured in FBM with 3.5 ϫ 10 5 cpm [ 14 C]G D1a /well in a 24-well plate for 18 h. Then the medium was removed, and the cells were washed twice with PBS, trypsinized, and harvested. The cell membranes were separated, and membrane-bound radioactivity was quantified by scintillation counting.

125
I-Labeled EGF Binding-Binding of 125 I-labeled EGF to whole cells was assessed by modified Scatchard for 18 h analysis (33,34). Three parallel sets of 10 4 NHDF were grown for 24 h in a 96-well plate in FGM-2, washed twice with FBM, and incubated with 0, 5, or 10 M G D1a in FBM for 18 h. One set of cells was then washed and incubated at 4°C for 2 h with 0.5-20 ng/ml 125 I-labeled human recombinant EGF in FBM without cold EGF to determine the total binding. The second set, used to test for nonspecific binding, was treated as above but with the addition of 300 ng/ml unlabeled EGF. After washing, 50 l of lysis buffer were added to each well, the samples were lysed on ice for 30 min, and the cpm of 125 I-labeled EGF in 20-l aliquots were counted. Specific binding was calculated by subtracting the nonspecific binding of 125 Ilabeled EGF from the total binding. The third set of cells was trypsinized, and the cell number was determined. The binding curves and Scatchard analysis of the data were performed using GraphPad Prism 3.03 software.
Assessment of EGFR Dimerization and Ligand-independent Dimerization-NHDF cells were cultured in 100 ϫ 20-mm culture dishes and preincubated with G D1a in FBM for 18 h, washed, treated with (for dimerization) or without (for ligand-independent dimerization) 2 ng of EGF/ml in FBM for 5 min, and washed twice with ice-cold PBS. 3 ml of PBS containing 1 mg/ml BS 3 were added, and the cells were incubated on ice for 30 min (35,36), washed twice with ice-cold PBS, and lysed in 1 ml of lysis buffer for 20 min on ice. Total EGFR was immunoprecipitated with the EGFR antibody. 500 g of lysate protein were loaded onto a 7.5% SDS-PAGE, and dimerization of the EGFR was detected by Western blotting using an anti-EGFR antibody.
Data Analysis-All Western blots shown are the results of a representative experiment of 2-4 separate experiments. The optical density of protein bands on the Western blots was quantified using software from the Scion Corp, and the mean Ϯ S.D. is shown in the bar graphs, with control levels set at 100 units. The -fold induction or percentage increase was calculated by comparing the differences of the means between control and treatment groups.

Enhancement of EGFR Activation by G D1a
Exposure of NHDF-Our previous findings that ganglioside preincubation of NHDF caused a concentration-dependent enhancement of EGF-induced fibroblast proliferation (25) and of EGFR autophosphorylation (19) led to the present investigations to determine how ganglioside enrichment of the cell membrane enhanced EGFR activity. In initial experiments we confirmed the effect of G D1a ganglioside on EGFR-mediated signaling in NHDF under the same conditions as those used in our previous studies (19). That is, after an 18-h preincubation of NHDF with 20 M G D1a in serum-free FBM, the cells were washed to remove unbound ganglioside and exposed to EGF (2 ng/ml) for 5 min in FBM. Under these conditions G D1a preincubation enhanced EGF-induced EGFR autophosphorylation by ϳ70% without affecting the total EGFR level (Fig. 1A). The specific inhibitor of EGFR autophosphorylation AG1478 (1.0 M), added during the last 3 h of the preincubation with G D1a , blocked both the EGF-induced and the G D1a -enhanced, EGFinduced EGFR autophosphorylation (Fig. 1A), demonstrating the specificity of the ganglioside effect on EGFR autophosphorylation. These overall findings were confirmed in three separate experiments (Fig. 1B).
A major EGF-induced signaling pathway is that of EGFR activation of Ras, which in turn activates Raf, MEK (mitogenactivated protein kinase/extracellular signal-regulated kinase kinase), and MAPK. Because MAPK plays an important role in cell proliferation (37), we examined the effect of G D1a on the activation of p44/42 MAPK. Under the same conditions that caused enhanced cell proliferation and enhanced EGFR autophosphorylation, 20 M G D1a preincubation enhanced the EGFinduced phosphorylation of p44/42 MAPK in NHDF by 65%, whereas the total MAPK level was unchanged. This increased MAPK phosphorylation was also inhibited by exposure of the cells to 1.0 M AG1478, further confirming the specificity of the ganglioside effect as being EGF-mediated ( Fig. 1C) as well as excluding a direct activating effect of G D1a downstream of the EGFR. Three additional separate experiments measuring the enhancing effect of G D1a and the inhibitory effect of AG1478 on MAPK phosphorylation (Fig. 1D) confirmed these results.
Enhanced EGFR Activation in Isolated NHDF Membranes by G D1a Preexposure-To establish whether the ganglioside effect was linked specifically to a membrane interaction, NHDF were preincubated with G D1a in FBM for 18 h as above, and then the plasma membranes were isolated. The isolated plasma membranes were exposed to EGF in FBM for 5 min and then lysed. The EGFRs were immunoprecipitated with the anti-EGFR antibody and analyzed by SDS-PAGE and Western blot using the phosphotyrosine antibody, Tyr(P) (PY99). As seen in the Western blot of a representative experiment ( the isolated membranes in a concentration-related manner, with a 91% increase in EGFR autophosphorylation in membranes of cells that had been preincubated with 20 M G D1a (Fig. 2B). The enhanced EGFR autophosphorylation found after G D1a enrichment both in intact cells and in isolated membranes localizes the ganglioside effect on EGFR phosphorylation to the plasma membrane.
Enhancement of EGFR-tyrosine Kinase Activity by G D1a Exposure-As a second experimental approach to confirm the enhancement of EGFR activation that we detected as enhanced autophosphorylation, we measured tyrosine kinase activity of the EGFR by measuring phosphorylation of a substrate peptide (38,39) after an 18-h preincubation of NHDF cells in 10 or 20 M G D1a and then a 5-min exposure to 2 ng/ml EGF. The EGFR tyrosine kinase activity in the intact cells (Fig. 3A) and in membranes isolated from these G D1a -pretreated cells (Fig. 3B) was quantified after immunoprecipitation of EGFR with the EGFR antibody. EGFR-tyrosine kinase activity was detected in NHDF after EGF stimulation, and ganglioside enrichment of the membrane enhanced this activity. After preincubation with 20 M G D1a , EGF-induced tyrosine kinase activity was increased both in whole cells (47%, Fig. 3A) and in isolated membranes (60%, Fig. 3B). Of note, G D1a itself did not enhance EGFR-tyrosine kinase activity, which excludes a direct activating effect of G D1a on EGFR signaling.
Plasma Membrane Binding of Exogenous [ 14 C]G D1a Ganglioside-To link the above enhancement directly to an effect of the ganglioside, we determined whether G D1a bound to the cell membrane. We exposed NHDF to 5 M [ 14 C]G D1a ganglioside for 18 h (as above), separated the plasma membranes, isolated and purified the membrane gangliosides, and quantified the membranebound radiolabeled G D1a . Approximately 0.15 Ϯ 0.01% of the added [ 14 C]G D1a was bound to the NHDF membranes, reflecting binding of 5.0 Ϯ 0.1 ϫ 10 6 molecules/cell, thus clearly documenting G D1a enrichment of the NHDF membranes.
Kinetics of EGFR Autophosphorylation-The dependence of enhanced EGFR autophosphorylation on the duration of G D1a preincubation was determined by incubating subconfluent NHDF were seeded as in Fig. 1 and preincubated with G D1a in FBM for 18 h. The cells were then treated with EGF (2 ng/ml) for 5 min and lysed, EGFR was immunoprecipitated, and EGFR protein-tyrosine kinase activity using an artificial substrate was quantified as described under "Experimental Procedures" (panel A). In panel B, the membranes of G D1a -preincubated cells were first isolated as in Fig. 2 and then treated with EGF (2 ng/ml) for 5 min, after which the EGFR were immunoprecipitated, and protein-tyrosine kinase activity was quantified. Protein-tyrosine kinase activity, expressed as net cpm of 32 P incorporated into the substrate peptide/100 g of cell lysate protein or 10 g of cell membrane lysate protein/10 min, was compared with that of untreated control cells; the mean Ϯ S.D. of -fold differences of three experiments are shown. NHDF for 5 min up to 6 h with 20 M G D1a in FBM. Then the cells were washed and incubated with EGF (2 ng/ml) for 5 min. Western blotting showed a substantial and maximal increase in EGFR autophosphorylation after only a 30-min pretreatment of the cells with G D1a (Fig. 4). Longer incubations maintained the same level of enhanced (ϳ2-fold) receptor autophosphorylation.
To assure that the enhancement did not reflect simply an alteration of the timing of the peak level of activation, we quantified the effect of G D1a preincubation on the time course of EGF stimulation (Fig. 5). NHDF cells were preincubated with 20 M G D1a in FBM for 18 h, washed, and then exposed to 2 ng/ml EGF in FBM for 5-30 min. Stimulation of the cells with EGF rapidly activated the EGFR (5 min), with a return to nearly base-line levels after 30 min. The magnitude (but not the time course) of activation was altered when cells were preincubated with G D1a ; EGFR phosphorylation was enhanced over the entire 30-min time course of EGFR activation, with the greatest enhancing effect observed at the peak of EGFR phosphorylation (5 min).
Ganglioside Modulation of 125 I-Labeled EGF Binding to NHDF-Because enhancement of EGF-induced fibroblast EGFR autophosphorylation by ganglioside G D1a is associated with ganglioside binding to the cell membrane, we tested whether membrane enrichment in G D1a affects binding of EGF to its receptor by quantifying concentration-dependent binding of 125 I-labeled EGF to NHDF at 4°C (40,41). As shown in Fig.  6 (inset), 125 I-labeled EGF binding to NHDF was nearly saturated at 10 ng/ml. G D1a preincubation greatly increased 125 Ilabeled EGF binding (p Ͻ 0.007). For example, after a 10 M G D1a preincubation, NHDF binding of 20 ng/ml 125 I-labeled EGF was increased 2-fold, from 0.34 pmol/10 8 cells (control) to 0.63 pmol/10 8 cells (p ϭ 0.0034). Scatchard analysis of the binding data (Fig. 6) confirmed that, as previously shown (40,42), more than one affinity class of EGFR exists in the membrane. Ganglioside enrichment by G D1a preincubation increased the total number of EGF molecules bound (e.g. from 3.0 ϫ 10 3 /cell to 4.6 ϫ 10 3 /cell after 10 M G D1a preincubation), affecting mainly high affinity binding (Table I). This included a striking 5-and 9-fold increase in the number of high affinity EGFR after 5 and 10 M G D1a preincubation, respectively. Furthermore, the proportion of high affinity EGFR was clearly increased, from 10% (control) to 58% (after 10 M G D1a preincubation) of total EGFR, suggesting a significant shift in receptor affinity for EGF (from low to high). Thus, these binding studies uncovered increased binding of 125 I-labeled EGF to NHDF, a somewhat increased total number of EGFR, and a striking increase in both the absolute number and proportion of high affinity EGFR, all without altering total receptor protein.
Ganglioside G D1a Enhances EGFR Dimerization-The significantly increased EGF binding after membrane ganglioside enrichment suggested further consideration of receptor interactions. EGFR exists both as monomers and as dimers. Because it is known that the high affinity form of the EGFR, which binds EGF more effectively, is the dimer (43), it seemed reasonable to suppose that increased dimerization might result from membrane ganglioside enrichment. To investigate this possibility, we determined whether G D1a preincubation affected EGF-induced EGFR dimerization. We detected EGFR dimers using the chemical cross-linker BS 3 (36), since BS 3 does not cross the plasma membrane and, therefore, detects only complexes that interact extracellularly. In five separate experiments, NHDF preincubated in G D1a for 18 h and treated with EGF for 5 min were subjected to cross-linking and then lysed. This was followed by immunoprecipitation with the EGFR antibody, Western blotting, and detection with the EGFR IgG (sheep polyclonal IgG) (Fig. 7). As shown in Fig. 7A, G D1a preincubation markedly increased the amount of EGFR dimerization. For example, 20 M G D1a pretreatment caused an approximate 60% increase in the amount of dimerization over that induced by EGF alone. These findings suggest that increased dimerization is a critical step in the ganglioside effect on EGFR signaling. However, possibly the most significant finding of these same experiments is that enrichment of the membrane with G D1a without exposure to EGF also caused markedly increased EGFR dimerization (Fig. 7B). That dimerization is a consequence of EGF binding to its receptor is well known, but recent studies have also identified the existence of ligand-independent EGFR dimerization in normal cells (44,45). In this case the dimer can be clearly detected, but receptor phosphorylation, evidencing activation, is not until EGF is added. Thus, EGFR dimerization and EGFR phosphorylation are separable processes (44,46,47). Our striking finding of FIG. 4. Effect of duration of G D1a preincubation on the enhancement of EGFR autophosphorylation. Subconfluent NHDF were washed twice with FBM and incubated with G D1a (20 M) in FBM for 0 -6 h or without G D1a for 6 h. The cells were washed, exposed to EGF (2 ng/ml) in FBM for 5 min, and harvested. EGFR were immunoprecipitated as before. In panel A, after SDS-PAGE and Western blotting, EGFR autophosphorylation was detected by an anti-phosphotyrosine antibody. Total EGFR was detected using an anti-EGFR sheep polyclonal IgG. In panel B, the composite findings (mean Ϯ S.D.) of three separate experiments are shown. FIG. 5. Effect of G D1a preincubation on the time course of EGFR autophosphorylation. NHDF were incubated with 20 M G D1a in serum-free FBM for 18 h, then exposed to EGF (2 ng/ml) for 0 -30 min followed by detection of EGFR phosphorylation as in Fig. 1. increased dimer formation by G D1a preexposure alone, almost to the level detected in the presence of an optimal concentration of EGF (Fig. 7), suggests that membrane enrichment in G D1a has a significant growth factor-independent effect on EGFR.
We further characterized this G D1a -induced ligand-independent EGFR dimerization by quantifying EGFR phosphorylation. Dimerization of normal EGFR observed in the absence of EGF might not be expected to initiate EGFR autophosphorylation (44,47). This is in fact what we found; EGFR autophosphorylation occurred only upon the addition of EGF and was enhanced by ganglioside preincubation of the cells (Fig. 8). In contrast, membrane enrichment in G D1a alone did not cause EGFR autophosphorylation (Fig. 8) but substantially increased EGFR dimerization (Fig. 7B). These findings are in fact consistent with those in Fig. 3, in which G D1a enrichment alone caused no stimulation of EGFR-tyrosine kinase activity. Thus, our results identify a highly specific and previously unde-scribed effect of G D1a ganglioside enrichment of the cell membrane, that of causing ligand-independent EGFR dimerization.
Effect of Other Gangliosides on EGFR Dimerization and Autophosphorylation-To assess the specificity of G D1a membrane enrichment in causing ligand-independent EGFR dimerization in human fibroblasts, in an additional series of four separate experiments we measured the effect of membrane enrichment in two other gangliosides, G M1 and G M3 , under the same conditions used in Fig. 7 and compared them to the effects of G D1a . When EFGR dimerization was determined in the absence of EGF, we found that G M1 enrichment resulted in a similar degree of dimerization as did G D1a enrichment (Fig. 9A). In contrast, G M3 enrichment was inactive in changing the degree of EGFR dimerization determined in the absence of ligand.
These findings were mirrored by the results obtained after EGF stimulation (Fig. 9A). Both G D1a and, albeit to a lesser degree, G M1 enhanced ligand-dependent dimerization, whereas  once again G M3 enrichment had no effect. Finally, EGFR autophosphorylation, the consequence of ligand-induced EGFR dimerization, was assessed in these same experiments to compare the effects of the three gangliosides. As shown in Fig. 9B, EGF-induced EGFR autophosphorylation after membrane enrichment in G D1a was nearly double that of control cells (p Ͻ 0.005). G M1 enrichment had a somewhat lesser, but still significant (p Ͻ 0.01) effect on receptor autophosphorylation after EGFR treatment, whereasG M3 enrichment did not significantly alter autophosphorylation of EGFR (p ϭ 0.16). Thus, we conclude that the propensity of membrane ganglioside enrichment to cause enhanced EGFR dimerization and subsequent EGFR autophosphorylation is at least partially related to the structure of the ganglioside with which the membrane has been enriched and, therefore, is not a structure-independent property. DISCUSSION EGF binds to its receptor, initiates cell signaling, and controls cell proliferation. EGFR activation is the key step in the signaling pathway, and the active form of EGFR is usually located in the low buoyant density domain of the cell membrane (48), where it can form complexes with downstream proteins. The complexes are enriched in gangliosides, other glycosphingolipids, sphingomyelin, and cholesterol (49 -51). Located in the cell membrane, gangliosides affect specialized cell functions such as immune responses, cell attachment to the extracellular matrix, cell movement, and the uptake of molecules from the cell surface (10).
There has been great interest in the modulation of cell signaling by gangliosides in recent years, with studies spanning a number of experimental systems, cell types, ganglioside species, and experimental conditions (21)(22)(23)(24)(25)(26)(27). Diverse opinions exist as to whether ganglioside effects are in general stimulatory (20,45,52) or inhibitory (22,23,(53)(54)(55), as both have been described. Inhibitory effects include G M3 inhibition of EGFR tyrosine phosphorylation (56) and the participation of G M3 in the pathological condition of insulin resistance, for example; insulin resistance induced in 3T3-L1 adipocytes by tumor necrosis factor-␣ was associated with increased G M3 ganglioside expression, whereas pharmacological depletion of G M3 prevented the tumor necrosis factor-␣-induced defect in insulindependent phosphorylation of the insulin receptor substrate, IRS-1 (57). G M3 has also been shown to have an inhibitory effect on EGFR autophosphorylation by binding directly to the purified extracellular domain of the EGFR (22,24). In contrast, G D1a and G M1 have been shown to have a positive effect on cell signaling in various systems (19, 20, 25, 52, 58 -60), including the recent report (20) that G M1 binding to FGF-2 is required for the mitogenic activity of the growth factor, suggesting that cell-associated G M1 may act as a functional co-receptor for FGF-2. These interesting and diverse findings of a number of laboratories underscore the importance of undertaking studies to elucidate mechanisms by which gangliosides affect cell signaling.
We previously reported that ganglioside enrichment of mouse and human fibroblasts, by preincubation with exogenous gangliosides, enhanced EGF-stimulated fibroblast proliferation (25). Conversely, cellular ganglioside depletion, by inhibition of glucosylceramide synthase and ganglioside synthesis, was inhibitory. Ganglioside G D1a enrichment of NHDF also enhanced EGFR-mediated signaling, evidenced as enhanced EGFR autophosphorylation and Ras/MAPK activation (19). We concluded that membrane enrichment of NHDF cells with G D1a increases cell proliferation in concert with enhancement of EGF-induced signaling.
Here we have found that the enhancing effect of G D1a preincubation on EGF-stimulated EGFR autophosphorylation was blocked by AG1478, a specific inhibitor of EGFR activation. Further supporting the concept that G D1a directly affects EGFR activation was the observation that AG1478 treatment also reduced the EGF-induced and G D1a -enhanced increase in downstream MAPK activation. That G D1a has its effect at the membrane level was shown by the enhanced EGF-induced EGFR autophosphorylation and EGFR-tyrosine kinase activity in cell membranes that were isolated after G D1a pretreatment of the cells. And, using radiolabeled G D1a , we confirmed that G D1a could bind directly to the plasma membrane.
These combined findings pointed to an effect of G D1a on the EGF/EGFR interaction at the membrane level. Consequently, we assessed binding of radiolabeled EGF to its receptor. Scatchard analysis demonstrated the known existence of two EGF receptor populations with different EGF binding affinities, high and low (43,44,61), and G D1a enrichment of the NHDF membrane markedly increased the number of high affinity receptors (e.g. a 9-fold increase after a 10 M G D1a preincubation) with only a small increase in total receptor number and FIG. 7. Effects of G D1a pretreatment on EGFR dimerization. Subconfluent NHDF cells were washed, preincubated with G D1a , and treated with EGF as in Fig. 1. Then the cells were washed twice with ice-cold PBS and incubated with 3 ml of PBS containing 1 mg/ml BS 3 for 30 min on ice, washed again, and lysed in 1 ml of lysis buffer for 20 min. Immunoprecipitation and detection of EGFR dimers were performed as described under "Experimental Procedures." The relative optical densities of the EGFR dimer bands, representing the composite (mean Ϯ S.D.) results of five separate experiments, are shown below the Western blot. Panel A, ganglioside preincubation followed by EGF exposure; panel B, ganglioside incubation only (no EGF exposure).
FIG. 8. Effects of G D1a pretreatment on EGFR autophosphorylation. Subconfluent NHDF were treated with 5-20 M G D1a in FBM for 18 h, exposed to 2 ng/ml EGF in FBM for 5 min, and lysed. Then, EGFR autophosphorylation (P-EGFR) was quantified as described in Fig. 1. none in total receptor protein. The difference between the receptor populations is that the high affinity receptor population consists of receptor dimers, whereas monomers constitute the low affinity receptor population (43,62). That the dimer population was markedly increased suggested that membrane ganglioside enrichment affects membrane topology in a way that facilitates receptor dimerization.
In the well known and generally accepted model of EGFR signaling, stable dimerization of the EGFR enhances its signaling, and EGFR dimerization follows EGF binding in normal cells (1). Restated, increased EGF binding precedes increased dimer formation (3,63) in turn followed by receptor autophosphorylation on intracellular tyrosine residues and receptortyrosine kinase activity. In our experiments the increases in high affinity receptor number, receptor dimerization, and receptor activation that followed G D1a enrichment of the membrane all would be consistent with this model, and they would suggest an EGF dependence of the modulatory effect of G D1a , i.e. that ganglioside enrichment followed by EGF exposure enhances binding of EGF to its receptor, which then rapidly leads to the higher affinity binding that characterizes the dimeric state (46). However, the completely unexpected finding of striking receptor dimerization caused by G D1a enrichment without exposure of the cell to the growth factor suggests an even more fundamental and a novel effect that membrane ganglioside enrichment causes self-association of EGFR into dimers. This interpretation suggests a new type of effect of gangliosides on cell signaling, that of stochastically increasing FIG. 9. Comparison of effects of G D1a , G M1 , and G M3 membrane enrichment on EGFR dimerization and autophosphorylation. Panel A, subconfluent NHDF cells were washed, preincubated with 20 M G D1a , G M1 , or G M3 treated with or without EGF, and EGFR dimerization was quantified by BS 3 crosslinking, immunoprecipitation, and detection as in Fig. 7. The relative optical densities of the EGFR dimer bands, representing the composite mean Ϯ S.D. of four separate experiments, are shown below the Western blot. Left, ganglioside incubation only (no EGF exposure). Right, ganglioside preincubation followed by EGF exposure. Panel B, subconfluent NHDF treated as above and preincubated with 20 M G D1a , G M1 , or G M3 in FBM for 18 h were exposed to 2 ng/ml EGF in FBM for 5 min and lysed. Then EGFR autophosphorylation (P-EGFR) was quantified as described in Fig. 1. The bar graphs represent the composite mean Ϯ S.D. of seven separate experiments. the proximity of these receptors to one another, thereby enhancing efficiency of binding and signaling once stimulated by the growth factor. The effect appears to be somewhat specific for ganglioside structure, since although GM1 ganglioside shared these properties with GD1a, GM3 ganglioside was essentially inactive.
These novel findings are in fact not inconsistent with a new perspective on receptor dimerization, i.e. that ligand-independent dimer formation of EGFR is a process separable from ligand-induced EGFR activation and signaling (44) and that EGF may act on preformed EGFR dimers to induce rotation of their transmembrane domains, which then permits intracellular kinase activation to occur (64). This contrasts to the more generally held view that EGFR dimerization requires and follows rather than precedes EGF binding to the receptor. Such ligand-independent dimerization, because it does not activate EGFR phosphorylation or subsequent signaling, has also been termed predimerization (44), and this is exactly what we found in ganglioside-enriched fibroblasts not exposed to EGF; that is, receptor dimerization without activation. Experimental strategies, which have been found to increase such preformed dimers, have included EGFR gene amplification or protein overexpression and the development of mutant EGFR constructs (65). However, to our knowledge the present findings are the first demonstration of a potential physiological/pathophysiological cause of substantial ligand-independent EGFR dimer formation in normal cells, and they illuminate the developing concept of ligand-independent dimerization of normal growth factor receptors.
In conclusion, because G D1a enrichment does not in itself trigger EGFR activation, we speculate that alteration of membrane topology by enrichment of the cell membrane in gangliosides such as G D1a may increase receptor-receptor proximity, triggering ligand-independent dimerization, in turn shifting the equilibrium between EGFR monomers and dimers toward dimers, resulting in increased sensitivity or responsiveness to EGF. Binding of EGF to its receptor is then dramatically increased, as seen by the Scatchard analysis, and is followed by significantly enhanced growth factor-dependent signaling and, as previously shown, increased fibroblast proliferation. A possible pathophysiologic cause of such a process, enrichment of the cellular microenvironment in exogenous gangliosides followed by binding of these molecules to normal cells (such as fibroblasts), causing enrichment of their membrane ganglioside content (66), is the rapid shedding of gangliosides by tumor cells (11,14,67,68).