Interaction of ganglioside GD3 with an EGF receptor sustains the self-renewal ability of mouse neural stem cells in vitro

Significance

Gangliosides are abundantly expressed in the nervous system and are known to play important roles in neurodevelopment. We previously demonstrated that GD3 is a predominant ganglioside species (>80% of the total gangliosides) in mouse neural stem cells (NSCs). To investigate its biological function, we found that GD3 is colocalized and interacts with the mitogen receptor EGFR in the microdomain structure of the plasma membrane. Such an interaction preserves the EGFR via an endosomal-plasma membrane recycling pathway after endocytosis of EGF and facilitates the EGF-mediated signaling responsible for regulating NSC cell-fate determination. This is an example of how a specific ganglioside can mediate the reuse of a mitogen receptor for promoting efficient cellular proliferation.

Abstract

Mounting evidence supports the notion that gangliosides serve regulatory roles in neurogenesis; little is known, however, about how these glycosphingolipids function in neural stem cell (NSC) fate determination. We previously demonstrated that ganglioside GD3 is a major species in embryonic mouse brain: more than 80% of the NSCs obtained by the neurosphere method express GD3. To investigate the functional role of GD3 in neurogenesis, we compared the properties of NSCs from GD3-synthase knockout (GD3S-KO) mice with those from their wild-type littermates. NSCs from GD3S-KO mice showed decreased self-renewal ability compared with those from the wild-type animals, and that decreased ability was accompanied by reduced expression of EGF receptor (EGFR) and an increased degradation rate of EGFR and EGF-induced ERK signaling. We also showed that EGFR switched from the low-density lipid raft fractions in wild-type NSCs to the high-density layers in the GD3S-KO NSCs. Immunochemical staining revealed colocalization of EGFR and GD3, and EGFR could be immunoprecipitated from the NSC lysate with an anti-GD3 antibody from the wild-type, but not from the GD3S-KO, mice. Tracking the localization of endocytosed EGFR with endocytosis pathway markers indicated that more EGFR in GD3S-KO NSCs translocated through the endosomal−lysosomal degradative pathway, rather than through the recycling pathway. Those findings support the idea that GD3 interacts with EGFR in the NSCs and that the interaction is responsible for sustaining the expression of EGFR and its downstream signaling to maintain the self-renewal capability of NSCs.

Gangliosides are ubiquitously expressed in all vertebrate cells and are particularly abundant in the nervous system (1). In early mammalian embryonic brain, the pattern of ganglioside expression is limited to simple gangliosides, predominantly GM3 (NeuAcα2-3Galβ1-4Glcβ1-1′Cer) and GD3 (NeuAcα2-8NeuAcα2-3Galβ1-4Glcβ1-1′Cer). In later developmental stages, however, more complex gangliosides prevail, particularly GM1 (Galβ1-3GalNAcβ1-4(NeuAcα2-3)Galβ1-4Glcβ1-1′Cer), GD1a (NeuAcα2-3Galβ1-3GalNAcβ1-4(NeuAcα2-3)Galβ1-4Glcβ1-1′Cer), GD1b (Galβ1-3GalNAcβ1-4(NeuAcα2-8NeuAcα2-3)Galβ1-4Glcβ1-1′Cer), and GT1b (NeuAcα2-3Galβ1-3GalNAcβ1-4(NeuAcα2-8NeuAcα2-3)Galβ1-4Glcβ1-1′Cer) (2, 3). Because of the spatiotemporal expression patterns, gangliosides abundant in embryonic brain, such as GD3 and the c-series ganglioside antigen marker A2B5, have been considered to be useful stage-specific markers of early brain development (4, 5). Mounting evidence supports the notion that gangliosides serve regulatory roles in neurogenesis through modulating processes, such as intercellular recognition, interaction, adhesion, reception, and/or signaling (6–9). Little is known, however, about how glycosphingolipids (GSLs) function in neural stem cell (NSC) fate determination. Investigation of the biological significance of gangliosides has also been greatly facilitated by the analysis of genetically engineered mice deficient in one or more ganglioside synthases (10). Most of the glycosyltransferase (GT)-knockout (KO) mice exhibit neural dysfunction and degenerative changes with aging. The mechanism of neural degeneration caused by the absence of gangliosides, however, has not been fully elucidated (11).

GD3 (CD60a), a b-series disialoganglioside, is known to be the dominant GSL species expressed in embryonic rodent brain, but its concentration rapidly decreases soon after birth (2). In adult mouse brain, GD3 was found to be primarily localized in the subventricular zone (SVZ) of the lateral ventricle, where NSCs robustly exist. In primary NSCs prepared from mice, the expression of GD3 was found to coexist with that of stage-specific embryonic antigen 1, another mouse NSC marker, in cultured neurospheres and neuroepithelial cells (NECs), but not in differentiated cells. For that reason, GD3 can be considered a useful NSC surface marker (4). The specific expression of GD3 in NSCs and its dynamic change during neural development prompted us to investigate the functional role of GD3 in regulating NSC fate determination.

Recently, a series of studies demonstrated that the activation and functional role of EGF receptor (EGFR) and β1-integrin in cell proliferation and migration were associated with GSL-enriched membrane domains, the so-called lipid rafts (12–14). Sorting of growth factor and signaling receptors, including EGFR intracellular trafficking, have been shown to be essential processes for the regulation of stem cell self-renewal (15). Other mechanisms, such as how EGFR signaling can be maintained in stem cells with continuous endocytic recycling, however, have not been evaluated.

Here we show that NSCs from GD3-synthase KO (GD3S-KO) mice exhibited a decreased self-renewal capability, which was accompanied by reduced levels of EGFR expression and accelerated rates of EGFR degradation after EGF stimulation. In addition, a reduction of nestin-positive cells was also found at the SVZ of GD3S-KO mice. By studying GD3 and EGFR interaction in NSCs, we demonstrated that the presence of GD3 contributed to the raft domain-associated localization of EGFR. Those observations provide convincing evidence that GD3 plays an important role in maintenance of NSCs self-renewal capability by preserving the EGFRs from degradation and by facilitating their recycling to the membrane surface.

Results

Accelerated Loss of Self-Renewal Ability in GD3-KO NSCs Is Accompanied by a Decreased Level of EGFR Expression.

To elucidate the role of GD3 in neurogenesis, we used the neurosphere culture method to compare the phenotype of NSCs from embryonic and postnatal GD3S-KO mice with their wild-type littermates. Immunocytochemical staining revealed that GD3 was colocalized with nestin in neurospheres prepared from GD3S^+/+ mice, but GD3 expression was not detected in neurospheres prepared from GD3S^−/− animals (Fig. 1A). Most interestingly, the NSCs of GD3S-KO mice showed the ability to form neurospheres in the first two passages, but reduced in number and size (Fig. 1B and Fig. S1 C and D). Moreover, that difference in neurosphere formation between GD3S^+/+ and GD3S^−/− mice was more obvious in NSCs prepared from postnatal mice, and the loss of self-renewal capability in the GD3S^−/− NSCs was even more obvious with further passages (Fig. 1B). Consistent with the in vitro data, nestin was found to be colocalized with GD3 at the SVZ in the 1-mo-old wild-type mouse brain, and the number of nestin-positive cells in the GD3S^−/− brain is less than in the GD3S^+/+ brain (Fig. S1 A and B). That observation implies that GD3 plays a greater role in the maintenance, rather than the generation, of NSCs.

Fig. 1.

Accelerated loss of self-renewal ability in GD3-KO NSCs is accompanied by a decreased EGFR expression level. (A) Immunofluorescence of GD3 and nestin stain in neurospheres from GD3S^+/+ and GD3S^−/− mice. (B) Quantification of sphere number formed from every 10,000 cells demonstrates a neurosphere-forming defect in GD3S^−/− NSCs. (C) Lysates (50 µg protein) of neurospheres from embryonic day (E) 14 embryos and P10 postnatal mice at passage 2 were immunoblotted with the indicated antibodies. (D) Lysates (50 µg protein) of monolayer NSCs growing in different growth factor conditions (20 ng/mL EGF and/or 20 ng/mL bFGF) for 48 h were immunoblotted with the indicated antibodies. (E) Lysates (50 µg protein) of neurospheres from E14 embryos with different passages were immunoblotted with EGFR. Data represent three independent experiments with three replicates of neurosphere pool for neurosphere number quantification. All values are expressed as mean ± SEM. *P < 0.05; **P < 0.01. (Scale bars, 50 µm.)

Previous studies demonstrated that cooperativity between the integrin and EGFR signaling pathways plays a crucial role in maintaining the embryonic neocortical stem cell niche and regulating both the NSC number and the ability of self-renewal (16, 17). To test the association of GD3 with integrin and EGFR, we measured the expression levels of EGFR and β1-integrin in neurospheres from GD3S^+/+ and GD3S^−/− mouse brains. We found reduced self-renewal ability in the GD3S^−/− NSCs, which was accompanied by a decreased expression of nestin and EGFR, but not of β1-integrin (Fig. 1C). The reduced proliferation rate was confirmed by the decrease of the MAPK/ERK pathway signaling in GD3S^−/− neurospheres (Fig. 1C). To test how the EGFR level was influenced in GD3S^−/− NSCs, primary neurospheres were dissociated into free suspension cells and were cultured as monolayer NECs in poly-l-ornithine (PLO)- and fibronectin-coated plates with different growth factor treatments. We found that GD3S-KO NSCs showed a noticeable decrease of EGFR expression in groups with EGF and with EGF plus basic fibroblast growth factor (bFGF) treatment, but not in the group with bFGF treatment alone (Fig. 1D). The expression of β1-integrin was not influenced under either condition. Further, the ability for the dissociated monolayer NSCs to reform secondary neurospheres was completely lost in GD3S-KO cells with low EGFR expression (the 20 ng/mL EGF treatment group) (Fig. S2A). Those results support a mitogen-dependent effect on the decrease of EGFR expression and the self-renewal defect in GD3S-KO NSCs. Moreover, we also found that the decrease in EGFR expression, which was accompanied by an accelerated loss of the self-renewal capability in the GD3S-KO NSCs, was even more pronounced with further passages (Fig. 1 B and E). At the same time, the mRNA level of EGFR showed no change between the GD3S^−/− and GD3S^+/+ NSCs from postnatal day (P) 10 mice at passage 3, both as neurospheres and NEC culture (Fig. S2 B and C). Taken together, those data clearly suggest that GD3 is required for maintaining the long-term self-renewal ability of NSCs by regulating the level of EGFR.

Accelerated Rates of EGFR and EGF-Induced ERK Signaling Degradation in NSCs Are Associated with GD3 Expression.

It is known that treatment of cells with EGF at high concentrations could induce EGFR degradation (18). To further examine how GD3 regulates the dynamics of EGFR expression, we monitored EGF-induced EGFR degradation and its downstream signaling in GD3S^+/+ and GD3S^−/− NSCs. Primary neurospheres were grown as monolayer NECs with 5 ng/mL EGF and 20 ng/mL bFGF for 16 h. EGFR degradation was induced by the addition of 50 ng/mL EGF stimulation for 0.25–3 h, whereas de novo protein synthesis was inhibited by cycloheximide treatment. In wild-type NSCs, no obvious EGFR degradation was detected within 3 h of EGF treatment, whereas in GD3S-KO NSCs, EGFR degradation and the reduction of p-EGFR and p-ERK1/2 signaling began at about 1 h (Fig. 2 A and B). We noticed that the p-AKT signaling did not show a corresponding decrease with the decrease of EGFR during the first 2 h but showed only a mild decrease at 3 h after the addition of EGF, suggesting that the MAPK/ERK proliferation pathway was the affected downstream pathway of EGF-induced signaling. Interestingly, β1-integrin expression did not decrease in GD3S^−/− NSCs but showed a mild increase with EGFR down-regulation (Fig. 2A).

Fig. 2.

Accelerated EGFR degradation in GD3S^−/− NSCs. NSCs were grown as monolayers in the presence of 5 ng/mL EGF and 20 ng/mL bFGF for 16 h. EGF degradation was induced by addition of 50 ng/mL EGF and 25 µg/mL cycloheximide at 37 °C for the indicated time. (A) Lysates (40 µg protein) were immunoblotted with antibodies against EGFR, p-EGFR, and β1-integrin. Densitometric analysis indicated that EGFR degradation was faster in GD3S^−/− NSCs than in GD3S^+/+ cells, whereas β1-integrin showed a slight increase with EGFR degradation. (B) Lysates (40 µg protein) were immunoblotted with antibodies against p-Akt, t-Akt, p-ERK1/2, or t-ERK1/2. Densitometric analysis revealed that EGF-induced p-ERK1/2 signaling is reduced in GD3S^−/− NSCs and that p-Akt signaling is not influenced in the same way. (C) Flow cytometry was used to analyze cell membrane surface expression of EGFR with 50 ng/mL EGF treatment at 0.5 h. Values are expressed as mean ± SEM. *P < 0.05; **P < 0.01.

Because de novo protein synthesis was inhibited by cycloheximide in this experiment and no obvious EGFR degradation was detected in the GD3S^+/+ NSC group, our observation strongly suggests there was a recycling pathway for EGFR to sustain proliferation signaling with EGF stimulation. To further test that possibility, the abundance of membrane EGFR with 50 ng/mL EGF stimulation at 0.5 h was studied by flow cytometry. Membrane EGFR expression was detected in 53.02 ± 0.01% of NSCs from P10 GD3S^+/+ mice, but only in 21.62 ± 2.48% of NSCs from P10 GD3S^−/− mice as cell surface–positive (Fig. 2C). Thus, our results support the hypothesis that GD3 is critical in sustaining the EGFR level in the membrane and in maintaining the downstream signaling for proliferation of NSCs with EGF stimulation. That regulatory effect could be achieved by facilitating EGFR recycling to the membrane, where the receptors are supposed to localize for the initiation of signaling induced by EGF.

To provide further evidence that GD3 was associated with EGFR, we examined whether the decreased EGFR level in GD3S-KO NSCs could be rescued by GD3S gene overexpression. GD3-synthase (GM3:α2,8-sialyltransferase, ST-II, Sial-T2) is responsible for catalyzing the biosynthesis of GD3 from GM3. Overexpression of the ST-II gene in cells was carried out by nucleofection, using a p-EGFP-N1 vector containing the ST-II fusion gene. After transfection, NSCs were seeded onto PLO- and fibronectin-coated plates as monolayers and cultured with 20 ng/mL EGF and 20 ng/mL bFGF. The expression of GFP was used as a transfection efficiency indicator, and we managed to achieve a >60% efficiency at 48 h after transfection. In the GD3S^−/− NSCs, only cells with efficient ST-II-GFP transfection could be immunostained with anti-GD3 monoclonal antibody (mAb) R24 (Fig. 3A). That finding indicated that the vector containing the ST-II-GFP fusion gene was able to restore GD3 synthesis in the GD3S-KO NSCs. When we examined the abundance of EGFR expression 48 h after transfection, we found an increased EGFR level in the GD3S-KO NSCs transfected with the ST-II-GFP fusion gene compared with those transfected with GFP vector only (Fig. 3 B and C). Our overall experiment provided convincing evidence for the rescuing effect of GD3 in restoring EGFR expression in GD3⁻ NSCs.

Fig. 3.

Rescue effect of GD3 expression on EGF level by exogenous transfection of ST-II-GFP. (A) NSCs transfected with GFP-vector or ST-II-GFP were grown as a monolayer culture and immunostained with an anti-GD3 antibody, mAb R24, at 48 h after transfection. In GD3S^−/− NSCs, only cells with efficiently transfected ST-II-GFP had GD3 expression. (B) EGFR expression level was detected in monolayer cultures of NSCs at 48 h after GFP-vector or ST-II-GFP transfection. 75 kD GFP represents the ST-II-GFP. (C) Densitometric analysis revealed an increase of EGFR in GD3S^−/− NSCs with ST-II-GFP transfection compared with cells with GFP-vector transfection. Values are expressed as mean ± SEM. *P < 0.05; **P < 0.01. (Scale bar, 10 µm.)

GD3 Interacts with EGFR and Plays a Role in Association of EGFR with the Lipid Rafts.

Colocalization and interaction of EGFR with gangliosides in cells have been reported (19). To study how EGFR was associated with GD3, we performed confocal microscopy to detect the localization of EGFR and GD3 in NSCs. Gangliosides are known to express mainly on the cell membrane surface, and their surface immunoreactivity can be washed out by a detergent because of their lipid nature (20). For that reason, we first attempted to stain NSC membrane with EGFR and GD3 without detergent permeabilization (Triton X-100 treatment). Under that condition, GD3 expression was detected on the membrane surface (Fig. S3A, Left, as shown by costaining with f-actin). Partial colocalization of EGFR and GD3 can be observed on the cell membrane surface at a low EGF concentration (5 ng/mL) in the GD3S^+/+ cells (Fig. 4A, row 1). In GD3S^−/− cells, no GD3 staining was detected, and EGFR showed a scattered punctate expression pattern (Fig. S3B, row 1). At a higher EGF concentration (20 ng/mL), EGFR and GD3 colocalization in the GD3S^+/+ NSCs became even more obvious (Fig. 4A, row 2), and EGFR tended to accumulate on the GD3-positive raft regions on the membrane. In contrast, in the GD3S^−/− NSCs, the detectable membrane EGFR under high concentrations of EGF was less obvious (Fig. S3B, row 2). To observe the internalized EGFR, we stained the detergent-permeabilized cells by treatment with 0.3% Triton X-100 for 20 min after fixation. In that case, the GD3 immunostain in GD3S^+/+ cells was detected inside the cell, but not on the plasma membrane (Fig. S3A, Right, as shown by costaining with f-actin). Moreover, a portion of the internalized EGFR under higher EGF culture conditions was found to be colocalized with GD3 (Fig. 4A, row 3). When p-EGFR and GD3 were coimmunostained in Triton X-100 permeabilized cells, most of the p-EGFR staining was found to be colocalized with the internalized GD3 (Fig. 4A, row 4). Those results provide convincing evidence of an interaction between GD3 and EGFR in NSCs. Colocalization of EGFR and p-EGFR with the internalized GD3 further supports the possibility that GD3 plays an important role for EGFR intracellular trafficking in NSCs and also provides a platform for the initiation of its downstream signaling.

Fig. 4.

GD3 interacts with EGFR and plays a role in EGFR localization in the lipid raft fractions. (A) Colocalization of EGFR with GD3 in GD3⁺ NSCs under different EGF concentrations, as indicated. Box-enclosed areas are magnified on right. (B) Immunoprecipitation of neurosphere lysates with mAb R24 and blotted with EGFR or β1-integrin. (C) Distribution of indicated proteins from E14 passage 2 GD3S^+/+ and GD3S^−/− neurosphere lysates in the lipid raft fractions prepared by a detergent-free method. (D) Densitometric analysis of EGFR and β1-integrin distribution in each fraction showed fraction shifts of EGFR, but not of β1-integrin in GD3S^−/− NSCs, compared with the GD3S^+/+ NSCs. (Scale bar, 5 µm.)

Because binding of EGFR with GM3-coated beads has been reported by previous studies (21, 22), a GSL-coated bead assay was also used to determine whether colocalization and coimmunoprecipitation of EGFR and GD3 represent a bona fide biochemical interaction. A Western blot with anti-EGFR antibody showed binding of EGFR from NSCs lysate in GD3S^+/+ cells with GD3- and GM3-coated polystyrene beads, but not with GM1-coated beads (Fig. S3C). Coimmunoprecipitation also showed that mAb R24-immunoprecipitated EGFR was detected in the GD3S^+/+ neurosphere lysate, but not in the GD3S^−/− lysate (Fig. 4B). In contrast, β1-integrin was not detected in immunoprecipitated lysates by mAb R24 from either GD3S^+/+ or GD3S^−/− NSCs.

EGFR has been reported to localize in the lipid raft fraction in many cell lines (23). Because GD3 has shown an interaction with EGFR in NSCs, as indicated by colocalization and coimmunoprecipitation experiments, we hypothesized that the absence of GD3 on the cell membrane could alter the association of EGFR with the lipid raft domains. We therefore compared the partitioning of EGFR into the lipid raft fraction by OptiPrep density gradient fractionation (24). Equal aliquots of each fraction of the gradient were analyzed by Western blotting to determine the distribution of EGFR. Flotillin 1 was used as a lipid raft marker, and transferrin receptor, a non–lipid raft plasma membrane marker, was used as a control. The change in EGFR distribution was reflected by the band shift from low-density fractions (F2–F4) in the GD3S^+/+ NSCs to higher-density fractions (F6 and F7) in the GD3S^−/− NSCs (Fig. 4 C and D), indicating a fraction switch of EGFR localization in the absence of GD3. Interestingly, we also detected a lipid raft fraction partitioning of β1-integrin in both GD3S^+/+ and GD3S^−/− NSCs; no fraction shift was found, however, for β1-integrin (Fig. 4 C and D). This finding indicates that the presence of GD3 in NSCs has a specific role on the localization and function of EGFR. Importantly, the distributions of flotillin 1 and transferrin receptor did not show significant differences for either GD3S^+/+ or GD3S^−/− NSCs, indicating that the effect of GD3 on EGFR was highly specific and did not occur as the result of a general membrane property change. Those findings provide convincing evidence that GD3 is crucial in the maintenance of self-renewal capability in NSCs by recruiting EGFR to the raft-associated domain to sustain the EGF-induced downstream signaling.

GD3 Influences Intracellular Trafficking of EGFR in NSCs.

Because the expression of EGFR was found to be reduced in GD3S-KO NSCs, we used biotinylated Alexa Fluor 488 EGF (EGF-488) to determine whether EGF internalization was affected. As expected, we found that GD3S-KO NSCs had consistently less EGF-488 internalization than did wild-type NSCs during the time course studied (Fig. S4A). Statistical analysis showed that a significantly lower proportion of GD3S-KO NSCs had internalized EGF-488 and also that GD3S-KO NSCs had significantly less internalized EGF-488 puncta in each of the cell populations during the time course studied (Fig. S4B).

After endocytosis, EGFR can be sorted to lysosomes for degradation or recruitment to the endosomes for recycling. Our results so far support the concept that GD3 could sustain EGF-induced signaling by recruiting EGFR on the raft-associated membrane domain and also inhibited degradation in the lysosomes. To further test that possibility, we tracked the intracellular localization of EGFR by coimmunostaining with the early endosomal (EEA1), lysosomal (LAMP 1), and recycling endosomal (Rab11) markers. We found that under normal culture conditions, EGFR showed increased colocalization with the lysosomal marker LAMP 1 as well as decreased colocalization with a marker of recycling endosomes, Rab11, in the GD3S-KO NSCs (Fig. 5A). The colocalization rate with the early endosomal marker, EEA1, also showed a mild decrease in the GD3S-KO NSCs (Fig. 5A). Consistent with those results, when the dynamics of EGFR endocytosis during the time course was studied, endocytosis of EGFR was delayed, as characterized by a reduced colocalization rate with EEA1 at the 15-min point (Fig. 5B and Fig. S5A). At the 30-min and 60-min points, the EGFR of GD3S^+/+ NSCs showed an increased colocalization rate with Rab11, and the GD3S^−/− NSCs showed an obvious increased colocalization rate of EGFR with LAMP 1 (Fig. 5B and Fig. S5 B and C).

Fig. 5.

GD3 regulates EGFR endocytosis. (A) Colocalization of EGFR with EEA1, LAMP 1, and Rab11 in monolayer NSCs under normal culture conditions (20 ng/mL EGF and 20 ng/mL bFGF). The colocalization coefficiency was measured by Zeiss LSM-enhanced colocalization software. At least 20 cells were measured for each condition in each experiment. (B) NSCs were grown as monolayers in the presence of 5 ng/mL EGF and 20 ng/mL bFGF for 16 h, and EGFR internalization was induced by addition of 50 ng/mL EGF. Colocalization coefficiency was measured by fixing cell coverslips and coimmunostaining EGFR with EEA1, LAMP 1, or Rab11 at the indicated time. Mander’s overlap coefficient, which indicates the actual overlap of two signals, was used to estimate the extent of colocalization of two channels. Values are expressed as mean ± SEM; n = 3. *P < 0.05; **P < 0.01. (Scale bar, 5 µm.)

Discussion

The spatiotemporal specific expression patterns of GD3 in the embryonic brain and stem cell–enriched SVZ of the postnatal and adult mice indicate a potential functional role of GD3 in neurogenesis during neural development. Our previous study demonstrated that when NSCs were sorted by GD3 antibody, GD3⁺ cells showed a higher efficiency in generating neurospheres than did GD3⁻ cells (4). In the current study, we consistently found reduced self-renewal ability of NSCs from GD3S-KO mice brains compared with those from their wild-type littermates. In a previous study, we showed that GD3, but not GD1b or GT1b, is the dominant ganglioside species in NSCs (20). This supports our contention that it is GD3, not GM3, GD1b, or GT1b, that plays a specific role in NSC self-renewal.

EGF is one of the essential mitogens for the derivation and propagation of neurosphere cultures. By sorting for high EGFR-expressing cell populations in NSCs derived from the postnatal and adult SVZ, the higher EGFR expression level was found to correlate consistently with a higher efficiency of neurosphere formation (25). Thus, it is likely that the sustained EGF-dependent signaling contributes to SVZ homeostasis and neurogenesis. Here, our findings indicate that GD3 is required for maintaining the long-term self-renewal ability of NSCs by regulating the level of EGFR. Overexpression of the GD3S gene in NSCs partially rescued the GD3S-KO phenotype by increasing the EGFR protein level in GD3S-KO NSCs, providing further proof for the correlation between GD3 expression and the EGFR level.

EGFR regulation by glycolipids has been studied by others. For example, Hakomori’s laboratory first reported that the interaction of N-linked GlcNAc termini of EGFR with the oligosaccharides portion of GM3 has an inhibitory effect on the EGFR tyrosine kinase (26). The NeuAc-lactose core structure of GM3 was proven to be essential for its binding with the extracellular domains of EGFR in an in vitro study (21). GD3 is generated from GM3 by addition of an α2–8-linked sialic acid to its single sialic acid residue. Because GD3 shares similar structural features with GM3, it is possible that the carbohydrate moiety might also be responsible for the interaction between GD3 and EGFR, although a hydrophobic interaction still cannot be ruled out with certainty in this case. In our study, we observed colocalization of EGFR with GD3 in NSCs on a microscopic level. Our finding that EGFR is associated with the raft domain is in agreement with previous studies when electron microscopy was used (27). Under higher concentrations of EGF, colocalization of EGFR with GD3 was even stronger; that phenomenon can be explained as the effect of an EGF-induced coalescence of lipid raft domains on the membrane structures (19). In that manner, the clustering of EGFR enhanced the signaling platform. On the basis of coimmunoprecipitation studies, EGFR can be precipitated by anti-GD3 antibody mAb R24 in the GD3S^+/+ neurosphere lysate. Although the precise mode of interaction between GD3 and EGFR has not been clarified, our data have clearly shown an interaction between EGFR and GD3, which may lead to the functional change of EGFR in NSCs.

Association of EGFR with lipid rafts has been suggested to have a profound effect on the function of EGFR by influencing ligand binding and tyrosine kinase activation (28); the potential role of raft domain on receptor trafficking and recycling, however, has never been elucidated. Although the GD3S-KO NSCs may accumulate GM3, which has been shown to inhibit tyrosine kinase activation by binding to EGFR (22), our findings in this study point to an alternative mechanism by which gangliosides may regulate that growth factor receptor activity.

A dividing fetal or adult NSC generates two daughter cells with symmetric or asymmetric distribution of EGFR, which leads to a differential cell fate/potentiality determination (29). Endocytosis-mediated receptor degradation is emerging as a key regulator of asymmetric/polarized distribution of signaling effectors within a cell or during cell division (30). In our study, we tracked intracellular EGFR localization with endocytosis pathway markers. Our results support the model (Fig. S6) that GD3 can antagonize EGFR degradation during NSC division in response to EGF stimulation, by directing receptor trafficking into the recycling pathway, thus bypassing lysosomes to evade degradation. In contrast, an increased probability of EGFR degradation as opposed to recycling in GD3S^−/− NSCs could enhance a difference in signaling between mother and daughter cells, such that the down-regulated EGFR in the daughter NSCs would lead to loss of the ability for self-renewal and further differentiation. That model thus represents a unique route for EGFR and its ligands to regulate self-renewal and determine the fate of NSCs.

In this study, we have also investigated the possibility of involvement of β1-integrin in the loss of self-renewal ability in GD3S^−/− NSCs. We did not find any difference in its expression in neurospheres or monolayer cultures of GD3S^+/+ and GD3S^−/− NSCs, nor did we find any fraction shift of β1-integrin in neurosphere lysates prepared from GD3S^−/− or GD3S^+/+ cells. That finding might indicate that the regulatory mechanism of β1-integrin in NSCs is independent of EGFR and, in this case, is not influenced by GD3 deficiency.

In conclusion, our present study has elucidated the role of GD3 in maintaining the self-renewal ability of NSCs by sustaining EGF-induced EGFR signaling. We have also provided direct evidence of the interaction of GD3 with EGFR in the lipid-raft microdomains of NSCs, and that interaction provides a mechanism to regulate the level of EGFR. Our data also suggest two mechanisms by which GD3 can regulate EGFR signaling to sustain NSC self-renewal: through the enhancement of EGFR clustering and initiating EGFR downstream signaling by recruiting EGFR localization in the lipid-raft domain, and through the maintenance of consistent EGF-induced self-renewal signaling by preserving the internalized EGFR from degradation and by facilitating the recycling of EGFR to the raft domain on the cell membrane surface.

Materials and Methods

Animals.

The GD3S KO mice and their wild-type littermates were kindly provided by the courtesy of Richard Proia [National Institute of Diabetes and Digestive and Kidney Diseases, National Institutes of Health (NIH), Bethesda, MD]. All experiments with mice were performed in compliance with the guidelines of the LAS and Institutional Animal Care and Use Committee of Georgia Regents University.

Antibodies and Reagents.

Anti-GD3 mAb R24 was kindly provided by Ken Lloyd (Memorial Sloan Kettering Cancer Center, New York, NY). For other commercialized antibodies and reagents, see SI Materials and Methods for details.

Neural Stem Cell Culture.

Neurospheres were prepared from embryonic mouse (E14) ganglionic eminence or postnatal mouse SVZs (P10), following an established protocol (see SI Materials and Methods for details).

Preparation of Detergent-Free Lipid Raft Fractions.

The procedure for preparing detergent-free fractions was modified from MacDonald and Pike (24) (see SI Materials and Methods for details).

For more information on flow cytometry, quantitative RT-PCR, GSL-coated bead assay, immunoprecipitation and immunoblotting, confocal microscopy, constructs and transfection, and statistical analysis, see SI Materials and Methods.