Fucoganglioside α-fucosyl(α-galactosyl)-G_M1: a novel member of lipid membrane microdomain components involved in PC12 cell neuritogenesis

Materials

The products purchased consisted of antibodies for SFKs and phospho-SFKs (pTyr⁴¹⁶) (Cell Signaling Technology), v-Src (Calbiochem), Fyn (Santa Cruz Biotechnology), Yes (BD Biosciences), neurofilament 160, phospho-TrkA (tropomyosin receptor kinase A) (pTyr⁴⁹⁰) and ERK1/2 (extracellular-signal-regulated kinase 1/2) (Sigma), TrkA and phospho-ERK1/2 (Upstate) and HRP (horseradish peroxidase) (Jackson ImmunoResearch Laboratories). Other products were HRP-conjugated CTB (cholera toxin B subunit) (Sigma), HRP-conjugated anti-mouse IgG (Vector Laboratories), Alexa Fluor® 488-conjugated CTB, Alexa Fluor® 488-conjugated anti-(mouse IgG) and Alexa Fluor® 594-conjugated anti-(rabbit IgG) (Molecular Probes), mouse 2.5S NGF (nerve growth factor) (Invitrogen), and protein kinase inhibitors for genistein [5,7-dihydroxy-3-(4-hydroxyphenyl)-4H-1-benzopyran-4-one; 4′,5,6-trihydroxy-isoflavone], PP2 {4-amino-5-(4-chlorophenyl)-7-(t-butyl)pyrazolo[3,4-d]-pyrimidine} and TrkA specific inhibitor [17] (Calbiochem).

Cell culture

Mouse myeloma cells and hybridomas were grown in DMEM (Dulbecco's modified Eagle's medium) containing 10% (v/v) FBS (fetal bovine serum). PC12 (pheochromocytoma) cells were cultured in DMEM supplemented with 7% FBS and 7% HS (horse serum) or in TIP (transferrin/insulin/progesterone) medium (DMEM supplemented with 5 μg/ml transferrin, 5 μg/ml insulin and 0.2 μM progesterone) at 37 °C in a 10% CO₂ humidified atmosphere.

Preparation of DIM (detergent-insoluble membrane) fraction

The DIM fraction was prepared as described previously [18,19], but with slight modifications. PC12 cells (4×10⁷ cells) were lysed with 1 ml of lysis buffer consisting of 1% (v/v) Triton X-100 and TN buffer (25 mM Tris/HCl, pH 7.5, 150 mM NaCl, 0.1 mM PMSF, 1 mM NHSO₃, 1 mM benzamidine, 2 mM CaCl₂, 5 mM MgCl₂, 0.15 mM spermine and 0.3 mM spermidine). Lysates were transferred into centrifuge tubes (14 mm×95 mm; Beckman Ultra Clear), kept on ice for 30 min, and then mixed with 1 ml of 80% sucrose in TN buffer. Sucrose solutions (5.5 ml of 30% sucrose and 3.5 ml of 5% sucrose) layered on to the lysates in 40% sucrose were centrifuged in a SW41Ti swinging rotor centrifuge (Beckman) at 4 °C and 39000 rev./min for 16 h. Light-scattering bands located at the 5% and 30% sucrose interface were collected as the DIM fraction.

Production of mAbs

A Balb/c mouse was immunized several times by intraperitoneal administration of the DIM fraction using the RIBI adjuvant system for 1.5 months and subsequently boosted with the antigen a month later. Collected spleen cells were fused with X63-Ag8.653 myeloma cells and maintained in a hypoxanthine/aminopterine/thymidine growth medium containing 10% hybridoma cloning factor (Igen) at 37 °C in a 10% CO₂ chamber. The culture supernatants of the hybrid cells were screened with the DIM fraction by ELISA. The 22 hybrid clones positive for DIM were obtained and subcloned further by the limiting dilution method.

Purification of PC12 gangliosides

The total lipid extracted from the freeze-dried PC12 cells (0.85 g) or rat brain tissue (0.1 g) with chloroform/methanol (2:1, v/v) and with chloroform/methanol/water (5:8:3, by vol.) was loaded on to a PBA (phenylboronate agarose) 60 column (12 mm×60 mm) (Millipore). The column was washed with 5 vol. of chloroform/methanol (2:1, v/v) and was eluted with chloroform/methanol/water (5:4:1, by vol.). PBA eluate was loaded on to an anion-exchange column (Q-Sepharose, 6.3 mm×250 mm), washed with chloroform/methanol/water (30:60:8, by vol.), and then eluted with a gradient from chloroform/methanol/water (30:60:8, by vol.) to chloroform/methanol/2 M aqueous sodium acetate (30:60:8, by vol.). Three GSL fractions were pooled as the neutral fraction, low-acidic fraction, termed G1, and high-acidic fraction, termed G2. The G1 fraction, which contains the major immunoreactive lipid, was dialysed against water, then applied to a Senshu Pak Aquasil HPLC column (4.6 mm×250 mm) (Senshu-kagaku) and eluted with a gradient from chloroform/methanol (7:3, v/v) to chloroform/methanol/water (5:4:0.5, by vol.). Three purified gangliosides were pooled as G1-1, G1-2 and G1-3 fractions. Immunoreactive lipid was present in the G1-3 fraction.

TLC immunostaining assay

TLC immunostaining was performed according to the method of Higashi et al. [20] with slight modifications. Standard lipids were applied to a plastic TLC plate, Polygram Sil G (Macherey-Nagel), and developed with solvent, as described in the instructions. The chromatogram was soaked for 1 h at room temperature (∼24 °C) in PBS, containing 1% egg albumin and 1% PVP (polyvinylpyrrolidone) and then incubated with primary antibodies in PBS containing 3% PVP at 4 °C overnight. After washing with PBS, the chromatograms were re-incubated with HRP-conjugated anti-(mouse IgG) in PBS containing 3% PVP at room temperature for 1 h. The chromatograms were then visualized with a 4-chloro-1-naphthol/hydrogen peroxide solution.

Sugar composition analysis

Methylglycosides were obtained by methanolysis of sample GSLs with 5% anhydrous methanolic HCl at 100 °C for 3 h. Fatty acid methyl esters were extracted with hexane, and the methylglycosides remaining in the methanolic solution were concentrated, N-acetylated and trimethylsilylated. The sugar compositions were determined using a Shimadzu GC-2014 instrument equipped with a capillary column CBP-1 (0.25 mm×25 m) by raising the temperature from 100 °C to 250 °C at a rate of 5 °C/min.

FAB (fast atom bombardment) mass analysis

FAB mass spectra were obtained using a JEOL JMS-HX110 high resolution mass spectrometer equipped with an FAB ion source and XMS computer system. Xenon as an ionizing gas was used at an accelerating voltage of 6 kV neutral beam. The sample was dissolved in 30 μl of DMSO. A 1 μl aliquot of the solution was applied on to a stainless steel folder (1 mm×4 mm), followed by the addition of approx. 2 μl of triethanolamine. The samples were subsequently analysed.

Glycosidase digestion

The sample ganglioside was sonicated in 50 mM phosphate buffer (pH 6.6). Enzyme (α1-3,6-galactosidase, 200 m-units) (Calbiochem) was added and incubated for 68 h at 37 °C. The product was separated by TLC, probed with HRP-conjugated CTB, and visualized with a 4-chloro-1-naphthol/hydrogen peroxide solution.

Morphological assessment of neurite outgrowth

Collagen (Cellmatrix Type 1C; Nitta Gelatin) was diluted with HCl solution (pH 3.0) to 0.3 mg/ml and coated on to plates or dishes. PC12 cells were inoculated on to the coated plates or dishes and cultured in the TIP medium at 37 °C overnight. After treatment with or without protein kinase inhibitors at 37 °C for 30 min, PC12 cells were cultured further in the TIP medium containing NGF and/or mAb PR#1 at 37 °C for 24 h. The morphological changes of the cells were observed by phase-contrast microscopy in vivo.

MTT [3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyl-2H-tetrazolium bromide] assay

PC12 cells (5×10³) were seeded with TIP medium in 96-well plates, incubated at 37 °C for 2 h, and then supplemented with PR#1 or serum. At days 0–4 of the culture, MTT assays (Chemicon) were performed, as described in the supplier's instructions. The growth of cells was quantified by assessing the reduction of MTT to formazan, measured as the absorbance at 590 nm.

Immunoblotting

Proteins were separated on polyacrylamide gels after dissolving in Tris/SDS sample buffer (Daiichi-kagaku) and then transferred on to PVDF membranes (Millipore). The blots were blocked with BlockAce (Dainihon-Seiyaku) and incubated with primary antibody and then with HRP-conjugated anti-(mouse IgG). Bands were visualized by an ECL® (enhanced chemiluminescence) system (GE Healthcare).

Immunocytochemistry

PC12 cells were fixed with 3% (w/v) paraformaldehyde in PBS for 20 min at room temperature, followed by incubation in blocking solution (2% BSA and 5% normal goat serum in PBS, pH 7.4). The cells were incubated with primary antibodies (1 μg/ml) for 1 h at room temperature, and then with fluorochrome-conjugated secondary antibodies for 1 h at room temperature. The immunolabelled cells were examined with a Zeiss LSM510 confocal microscope.

Co-immunoprecipitation analysis

PC12 cells were lysed with lysis buffer [20 mM Tris/HCl, pH 7.4, 2 mM EDTA, 2 mM Na₃VO₄, 2 mM NaF, 0.5% Triton X-100 and 1% protease inhibitor cocktail (Sigma)] at room temperature, then incubated with PR#1 for 1 h at 4 °C. The immunocomplexes were precipitated after incubation with Protein G–agarose beads (KPL) for 1 h at 4 °C. The beads were washed several times with PBS. The immunocomplexes were separated by SDS/PAGE and analysed by immunoblotting, as described above. With HRP-conjugated CTB and anti-HRP antibody, G_M1a-associated proteins were analysed in the same way.

Analysis of tyrosine-phosphorylated proteins

PC12 cells (10⁶) were seeded on to collagen-coated 60-mm-diameter dishes. The cells were stimulated with 20 μg/ml PR#1, 10 ng/ml NGF or 1 μg/ml CTB, and incubated at 37 °C for the indicated times. NGF was used as a positive control in this assay. After the indicated periods, the cells were rinsed once with DMEM, lysed in Tris/SDS sample buffer containing 2 mM EDTA, 2 mM Na₃VO₄, 2 mM NaF and 1% protease inhibitor cocktail, and immediately boiled for 5 min. The lysates were then analysed by immunoblotting, as described above.

Identification of a novel ganglioside involved in neuritogenesis

The 22 hybrid clones obtained were grouped into four types, based on the Western blotting patterns. mAbs in one group recognized a glycolipid band, whereas others were protein bands or nonspecific broad bands non-specifically. mAbs recognizing glycolipids in the present study were subjected to the following treatment: following PC12 cell culture (5×10⁴ cells) for 24 h in the medium containing 10 μg/ml of each mAb from supernatant of the selected clones, mAb PR#1 (subclass IgG3) was found to have biological effects inducing neurite outgrowth on PC12 cells. No such effects were found in PC12 cells treated with other mAbs.

In order to identify the antigen of PR#1, a TLC immunostaining assay with total lipids extracted from PC12 cells or rat brains was carried out (Figure 1A). The immunoreactive lipid was alkali-resistant and detectable with orcinol and resorcinol reagent, indicating that the antigen of PR#1 could be a ganglioside, which was clearly localized in the DIM fraction (Figure 1B). Furthermore, the R_F value of the major PR#1 immunoreactant was located between the G_M1a and G_D1a, a region in which three gangliosides such as G_M1a, Fuc-G_M1 (IV²Fucα, II³NeuAc,GgOse₄Cer) and Fuc(Gal)-G_M1 (IV²Fucα,IV³Galα,II³NeuAc,GgOse₄Cer) have been reported to be present [21]. To specify the antigen, TLC/immunostaining with PR#1, CTB, which specifically reacted to G_M1a and weakly cross-reacted to Fuc-G_M1 [22], and the anti-G_M1a mAb DIM24 [23] was carried out on the purified ganglioside fractions obtained from the PC12 cell extract (see the Materials and methods section and Figure 2). Since CTB reacted with the gangliosides in fractions G1-1 and G1-2 and DIM24 with fraction G1-1, gangliosides in G1-1 and G1-2 were determined to be G_M1a and Fuc-G_M1 respectively (Figure 2A). The ganglioside in G1-2 was confirmed as Fuc-G_M1 by mass analyses as well (results not shown).

The G1-3 fraction contained a dense band of antigenic gangliosides of PR#1 (Figure 2A), and it remains to be confirmed whether this band is composed of a remaining candidate ganglioside Fuc(Gal)-G_M1. The hexose composition analysis showed that the ganglioside in G1-3 contains fucose, glucose, galactose, N-acetylgalactosamine and N-acetylneuraminic acid in the molar ratio 1:1:3:1:1 (Figure 2B). Figure 2(C) shows the negative-ion FAB mass spectrum and the fragmentation diagrams of the ganglioside. This spectrum showed the prominent molecular ions (M−H)⁻ m/z 1826, 1854, 1882, 1910 and 1938, which corresponded to the molecular species of Fuc(Gal)-G_M1 containing C_16:0, C_18:0, C_20:0, C_22:0 and C_24:0 fatty acids respectively. The ion groups a, b, b+sialic acid, c and d, corresponding to ceramide-bearing oligosaccharide fragments (699, 861, 1151, 1355 and 1664 correspond to ceramide-bearing oligosaccharide with C_16:0-sphingosine), and the ion A, corresponding to oligosaccharide residues, were also observed (Figure 2C). In addition, the digestion product of G1-3 ganglioside with α1-3,6-galactosidase was Fuc-G_M1 and was detected by its specific ligand CTB (Figure 2D). These data substantiate that the antigen of PR#1 is the fucoganglioside Fuc(Gal)-G_M1 (Figure 2E). Although the fraction G1-3 contained a small amount of phosphatidylinositol, PR#1 did not react with it (results not shown).

Cross-linking-like effects of Fuc(Gal)-G_M1 is distinct from G_M1a

Figure 3(A) shows the dose-dependence of PR#1 for the neurite outgrowth effect. The ratio of neurite-bearing cell number to total cell number increased with the concentration of PR#1 up to 20 μg/ml. The plateau value of the ratio (approx. 25%) in PR#1 treatment was approximately half that found in the NGF treatment. Effects on cell proliferation and viability, as assayed following the MTT protocol, were almost the same between PR#1-treated and non-treated cells, even when high doses of the antibody were added (Figure 3B). In PR#1-treated cells, aggregations of antigenic lipids were ubiquitously observed at the plasma membrane of cell bodies and neurites (Figure 3C), and most cells bearing neurites possessed such aggregations. These observations suggested that the cross-linking-like effect (if not ‘cross-linking’ in a strict sense) of Fuc(Gal)-G_M1 with PR#1 causes PC12 neurite outgrowth.

Since ganglioside G_M1a is known to form microdomains at the surface of the plasma membrane [24] and is involved in neuron-like differentiation of the PC12 cells [25–31], it is worthwhile to clarify whether Fuc(Gal)-G_M1-enriched domains, a homologous molecule, are similar to or distinct from G_M1a-formed domains. Using PR#1 and CTB, the morphological changes of PC12 cells induced by them were compared. Under PR#1 treatment for Fuc(Gal)-G_M1, the length of the cell processes continued to increase for 2–3 days, but no more elongation or morphological changes occurred thereafter (Figure 4, top row). The shape of the cell body transformed slightly into a rounded form. CTB treatment resulted in the development of cell processes longer than those generated in PR#1-treated cells and a cell soma with a flattened morphology, similar to that found with NGF treatment (Figure 4, second row from top). In both cases, the morphology of cells at later differentiation stages was distinct from that of NGF-treated cells that showed much longer processes and a round soma (Figure 4, middle row). Synergistic effects on cell differentiation by co-treatment of PR#1 and NGF were observed, i.e. much longer, multiple and profuse neurite branches in the case of addition of PR#1 (Figure 4, second row from bottom and Figure 5). In addition, this treatment led to faster development of roundness in the cell soma within 1 day, although it took at least 4 days in an ordinary NGF-induced differentiation (Figure 4, middle row), indicating that PR#1 promoted neuron-like differentiation. On the other hand, co-treatment with CTB and NGF induced mostly opposite effects in terms of the following: enlargement of cell bodies and neurites, including a subset of cells with a much more flattened form (Figure 4, bottom row).

Considering some contribution of neurofilaments in the cross-linking-like effects accompanying such morphological changes, the expression levels of neurofilament proteins were examined (cf. [32]). As shown in Figure 6, neither PR#1 nor CTB could induce NF-M (neurofilament M) expression, indicating that they were responsible for cell process outgrowth, but not for neural differentiation itself. On the other hand, co-treatment with NGF made a difference in terms of NF-M expression, i.e. additional PR#1 led to a slightly stronger expression than NGF treatment alone, whereas that of CTB resulted in a cancelling out of the expression, despite enhancing cell process outgrowth. These results indicate that the mechanisms underlying the cross-linking-like effects through Fuc(Gal)-G_M1 are distinct from those occurring through G_M1a.

Subcellular distribution of Fuc(Gal)-G_M1 and G_M1a

Using PR#1 and CTB, the distribution of Fuc(Gal)-G_M1 and G_M1a was compared immunocytochemically during the neuron-like differentiation of PC12 cells. There were three types of cells, i.e. G_M1a-dominant cells, Fuc(Gal)-G_M1-dominant cells and the cells expressing both of the gangliosides. These double-positive cells were used in the following analysis.

In undifferentiated cells, the majority of Fuc(Gal)-G_M1 immunoreactivity appeared as punctate structures in the soma, and G_M1a immunoreactivity similarly and partially overlapped with Fuc(Gal)-G_M1 signals (Figure 7, panels 1–3). These ganglioside signals never overlapped with the fluorochrome signals for mitochondria or lysosomes throughout the differentiation process (results not shown). At 2 days after NGF treatment (Figure 7, panels 4–6), Fuc(Gal)-G_M1 signals were found, in addition to the soma, to be concentrated at the tip portion of the extended neurites, whereas G_M1a signals remained in the soma. At 10 days after NGF treatment (Figure 7, panels 7–9), Fuc(Gal)-G_M1 signals at the neurite tips became less dense, but remained strongly confined to the soma. In contrast, G_M1a signals tended to be accumulated mainly on the cell surface of the soma and proximal portion of neurites, rather than in the cytoplasm. The behaviour of Fuc(Gal)-G_M1 in neuritogenesis was suggested to be significantly different from that of G_M1a throughout the differentiation process.

Certain SFKs are involved in signal transduction via the Fuc(Gal)-G_M1-enriched platform

In order to search for Fuc(Gal)-G_M1-associated proteins, the proteins were immunoprecipitated from Triton X-100 extracts of PC12 cells, using mAb PR#1 (Figure 8A). Identified proteins were certain SFKs, i.e. Fyn and Yes, but not Src itself. TrkA and Ras proteins, which have been known to be NGF-responsive receptors associated with G_M1a-enriched platforms [27] and the TrkA-signalling pathway respectively, were not detected. Fuc(Gal)-G_M1 was co-localized immunocytochemically in PC12 cells and co-precipitated with activated SFKs after 30 min of treatment with PR#1 (Figures 8B and 8C), in contrast with the TrkA with weak Yes signals in immunoprecipitates treated with CTB for 30 min (Figure 8A). These observations strongly suggest that SFKs other than Src are closely associated with Fuc(Gal)-G_M1 in PC12 cells.

The contribution of SFKs in signal transduction via Fuc(Gal)-G_M1 was confirmed by their phosphorylation specificity. Figure 9(A) shows the time course of protein tyrosine phosphorylation after PR#1, CTB or NGF treatment. In PR#1 treatment, an increase of tyrosine phosphorylation of SFKs was observed within 15 min, but was not seen with TrkA and ERKs. In CTB treatment, ERKs were activated strongly after 60 min and TrkA weakly within 5 min, but SFKs were not activated. Supportive results were obtained from examination of the effects of protein kinase inhibitors on the outgrowth of cell processes (Figure 9B). In this assay, the following inhibitors were used: (i) PP2, a specific inhibitor of SFKs; (ii) TrkA-specific inhibitor; and (iii) genistein, a wide-range tyrosine kinase inhibitor. After pre-treatment with PP2, cell process outgrowth induced by the addition of PR#1 was found to be strongly suppressed, whereas outgrowth by the addition of CTB and NGF were weakly or not at all suppressed. PP2 pre-treatment also suppressed the synergistic effect of PR#1 on neurite sprouting and branching of cells differentiated by NGF (results not shown). On the contrary, pre-treatment with the TrkA inhibitor suppressed process outgrowth attributable to CTB and NGF treatment, but no such effects were seen with PR#1 treatment. With genistein pre-treatment, process outgrowth was suppressed under all conditions. From the above findings, it was clear that PR#1-induced morphological changes were mediated exclusively by SFKs and independent of TrkA signalling pathways, which were involved in CTB and NGF activation of cell differentiation. In other words, a signal platform provided by Fuc(Gal)-G_M1 was found to be specifically associated with SFKs, distinct from a G_M1a platform which was associated with TrkA.