Immunoseparation of sphingolipid-enriched membrane domains enriched in Src family protein tyrosine kinases and in the neuronal adhesion molecule TAG-1 by anti-GD3 ganglioside monoclonal antibody

Authors

  • Alessandro Prinetti,

    1. Study Center for the Functional Biochemistry and Biotechnology of Sphingolipids, Department of Medical Chemistry and Biochemistry – Interdisciplinary Laboratory of Advanced Technologies, University of Milano, Segrate, Milan, Italy
    Search for more papers by this author
  • Simona Prioni,

    1. Study Center for the Functional Biochemistry and Biotechnology of Sphingolipids, Department of Medical Chemistry and Biochemistry – Interdisciplinary Laboratory of Advanced Technologies, University of Milano, Segrate, Milan, Italy
    Search for more papers by this author
  • Vanna Chigorno,

    1. Study Center for the Functional Biochemistry and Biotechnology of Sphingolipids, Department of Medical Chemistry and Biochemistry – Interdisciplinary Laboratory of Advanced Technologies, University of Milano, Segrate, Milan, Italy
    Search for more papers by this author
  • Domna Karagogeos,

    1. Institute of Molecular Biology and Biotechnology, and University of Crete Medical School, Crete, Greece
    Search for more papers by this author
  • Guido Tettamanti,

    1. Study Center for the Functional Biochemistry and Biotechnology of Sphingolipids, Department of Medical Chemistry and Biochemistry – Interdisciplinary Laboratory of Advanced Technologies, University of Milano, Segrate, Milan, Italy
    Search for more papers by this author
  • Sandro Sonnino

    1. Study Center for the Functional Biochemistry and Biotechnology of Sphingolipids, Department of Medical Chemistry and Biochemistry – Interdisciplinary Laboratory of Advanced Technologies, University of Milano, Segrate, Milan, Italy
    Search for more papers by this author

Address correspondence and reprint requests to Prof. Sandro Sonnino, Dipartimento di Chimica e Biochimica Medica – LITA-Segrate, Via Fratelli Cervi 93, 20090 Segrate, Milano, Italy. E-mail: Sandro.Sonnino@unimi.it

Abstract

Rat cerebellar granule cells differentiated in culture were fed [1–3H]sphingosine, allowing the metabolic radiolabelling of all cell sphingolipids and phosphatidylethanolamine. A detergent-insoluble sphingolipid-enriched membrane fraction, containing about 60% of cell sphingolipids, but only trace amounts of phosphatidylethanolamine, was prepared from [1–3H]sphingosine-fed cells by sucrose gradient centrifugation. This fraction was enriched in the Src family protein tyrosine kinases c-Src, Lyn and Fyn and in the GPI-anchored neuronal adhesion molecule TAG-1. The cell lysate and the sphingolipid-enriched membrane fraction were subjected to immunoprecipitation with anti-GD3 ganglioside monoclonal antibody R24, under experimental conditions designed to preserve the integrity of the domain. The radioactive lipid composition of the immunoprecipitates obtained from the cell lysate and from the sphingolipid-enriched fraction were very similar, and closely resembled the sphingolipid composition of the whole sphingolipid-enriched membrane fraction. In fact, the immunoprecipitates contained, together with GD3 ganglioside, all cell glycosphingolipids and sphingomyelin, whereas they did not contain phosphatidylethanolamine. Moreover, cholesterol and phosphatidylcholine were detected in the immunoprecipitates by qualitative TLC analysis followed by colourimetric visualization. c-Src, Lyn, Fyn and TAG-1 were associated with the anti-GD3 antibody immunoprecipitate. These proteins were not detected in the immunoprecipitates obtained under experimental conditions different from those designed to preserve the integrity of the domain. These data suggest that a membrane domain containing cholesterol, phosphatidylcholine, sphingolipids and proteins can be separated from the total cell membranes by anti-GD3 antibody immunoprecipitation, and that the association of c-Src, Fyn, Lyn, and TAG-1 with the sphingolipid-enriched domain is mediated by the interaction with a complex lipid environment, rather than by specific interactions with a single sphingolipid species.

Abbreviations used
GPI

glycosyl phosphatidyl inositol

HRP

horseradish peroxidase

IP

immunoprecipitation

PAGE

polyacrylamide gel electrophoresis

PNS

post-nuclear fraction

SDS

sodium dodecyl sulfate

SEMF

sphingolipid-enriched membrane fraction.

Cell membrane lipid domains, i.e. areas with a peculiar composition, different from that of the bulk membrane environment, progressively attracted the interest of scientists, following the hypothesis that they might participate in important functional membrane events such as signal transmission, cell adhesion and lipid/protein sorting (Hakomori 2000). Lipid domains enriched in sphingolipids and cholesterol, and containing phosphatidylcholine as the main glycerophospholipid (Prinetti et al. 2000a, 2001), have been separated from the rest of the cell components by procedures based on their resistance to solubilization (Smart et al. 1995; Rodgers and Rose 1996; Song et al. 1996) and their low buoyant density. Pictures of membrane domains enriched in glycosphingolipids have been obtained by electron microscopy through proteins that specifically bind gangliosides (Parton 1994; Chigorno et al. 2000).

The association of gangliosides with proteins in the sphingolipid-enriched membrane domains has been described. Among those, the interaction between GD3 and the Src-family tyrosine kinase Lyn has been proposed to affect the distribution of the glycosyl phosphatidyl inositol (GPI)-anchored adhesion molecule TAG-1 in the lipid membrane domain. GD3 ganglioside and Lyn were co-immunoprecipitated from rat cerebellar granule cells in culture by anti-GD3 antibody (Kasahara et al. 1997, 2000). This result suggested the existence of a specific interaction between GD3 and Lyn. Lyn is a myristoylated protein, associated with the membrane inner lipid layer. Thus, its interaction with GD3 ganglioside, belonging to the outer layer of the membrane, is likely mediated by hydrophobic interactions involving the myristoyl chain and the ceramide moiety of the ganglioside, rather than the ganglioside disialyl chain protruding in the extracellular environment.

In this work we show that a complex lipid environment, containing gangliosides, sphingomyelin, cholesterol, phosphatidylcholine and proteins, resembling the sphingolipid-enriched membrane domain, is immunoprecipitated by an anti-GD3 monoclonal antibody.

Materials and methods

Anti-Lyn, anti-Fyn, anti-c-Src (SRC2) rabbit polyclonal IgG antibodies and horseradish peroxidase (HRP)-conjugated secondary antibodies were from Santa Cruz Biotechnology (Santa Cruz, CA, USA). Anti-GD3 purified mouse monoclonal R24 antibody (Real et al. 1985) was obtained from the National Cancer Institute Biological Resources Branch Preclinical Repository (Rockville, MD, USA). Rabbit polyclonal antibodies to TAG-1 were raised against baculovirus-produced protein. Protein G-coupled magnetic beads (Dynabeads Protein G) were purchased from Dynal Biotech (Smestad, Oslo, Norway). [1-3H]Sphingosine (2 Ci/mmol) was prepared according to Toyokuni et al. (1991), and 3H-labelled lipids, used as chromatographic standards, were prepared according to Chigorno et al. (2000). Granule cells were obtained from the cerebellum of 8-day-old Sprague–Dawley rats and cultured as described in Gallo et al. (1982). Cells at the eigth day in culture were incubated in the presence of 3 × 10−8 m[1–3H]sphingosine (5 mL/dish) in cell-conditioned medium for a 2-h pulse followed by a 48-h chase (Prinetti et al. 2001). Sphingolipid-enriched membrane fractions from [1-3H]sphingosine-fed cells were prepared following the Triton X-100 method (Rodgers et al. 1996), modified as described in Prinetti et al. (2000a).

Aliquots of the post-nuclear fraction (PNS) and of the sphingolipid-enriched membrane fraction (SEMF) obtained from [1–3H]sphingosine-labelled cells (3.0 × 107 cells) were diluted 10-fold in immunoprecipitation (IP) buffer (50 mm Tris-HCl, pH 7.4, 150 mm NaCl, 2 mm NaF, 1 mm EDTA, 1 mm EGTA, 1 mm Na3VO4, 1 mm PMSF, 75 mU/mL aprotinin, 1% Triton X-100). After preclearing for non-specific binding, 12.5 µg/mL anti-GD3 mouse antibody R24 or 12.5 µg/mL normal mouse IgG (as negative control) was added to the supernatants, and the mixtures were stirred overnight at 4°C. Immunoprecipitates were recovered using protein G-coupled magnetic beads (Dynabeads Protein G), washed three times with IP buffer, recovered by centrifugation, and suspended in sodium dodecyl sulfate (SDS)-sample buffer. The radioactivity associated with immunoprecipitates was determined by liquid scintillation counting.

In some experiments, aliquots of the PNS were diluted in the above IP buffer and incubated at 37°C for 20 min to allow the disaggregation of the sphingolipid-enriched domains (Brown and Rose 1992) before performing the immunoprecipitation as described above. Alternatively, the immunoprecipitation experiments were performed using RIPA buffer (50 mm Tris-HCl, pH 7.4, 150 mm NaCl, 2 mm NaF, 1 mm EDTA, 1 mm EGTA, 1 mm Na3VO4, 1 mm PMSF, 75 mU/mL aprotinin, 1% Triton X-100, 0.2% SDS, 0.5% sodium deoxycholate) instead of the above IP buffer (Prinetti et al. 2000b).

Cell fractions, immunoprecipitates and supernatants remaining after immunoprecipitation were dialysed, lyophilized and lipids were extracted with chloroform/methanol 2 : 1 v/v (Prinetti et al. 2000a). The total lipid extract was analysed by HPTLC with the solvent systems chloroform/methanol/0.2% aqueous CaCl2, 50 : 42 : 11 or 55 : 45 : 10 v/v, followed by radioactivity imaging (Beta-Imager 2000 Z instrument; Biospace, San Francisco, CA, USA). Cholesterol was separated by monodimensional HPTLC using the solvent system hexane/diethylether/acetic acid, 80 : 20 : 1 v/v, and visualized with 15% concentrated sulphuric acid in 1-butanol. Phosphatidylcholine was separated by a two-run monodimensional HPTLC using the solvent system chloroform/methanol, 9 : 1 v/v, followed by the solvent system chloroform/methanol/acetone/acetic acid/water, 10 : 2 : 4 : 2 : 1 v/v. Proteins in PNS and fractions were separated by SDS–polyacrylamide gel electrophoresis (PAGE). After separation, proteins were transferred to polyvinylidene difluoride (PVDF) membranes and the presence of c-Src, Fyn, Lyn and TAG-1 was assessed by immunoblotting with specific antibodies, followed by reaction with secondary HRP-conjugated antibody and enhanced chemiluminescence detection (ECL, Pierce Supersignal). Protein content was determined (Lowry et al. 1951) using bovine serum albumin as a reference standard.

Results and discussion

It has been previously shown that lipid domains enriched in sphingolipids (gangliosides, neutral glycosphingolipids, ceramide and sphingomyelin) and cholesterol exist within the cell membrane. Recently, the lipid quantitative and qualitative composition of sphingolipid-enriched membrane domains has been studied in detail in neuronal cell cultures using metabolic radiolabelling procedures for the determination of sphingolipids and glycerophospholipids (Prinetti et al. 2000a, 2001). The use of [1-3H]sphingosine to radiolabel sphingolipids gave the advantage that phosphatidylethanolamine would also be radiolabelled, as a result of the recycling of the catabolic fragment ethanolamine. By this approach, we were able to demonstrate that the sphingolipid-enriched domains are devoid of phosphatidylethanolamine.

Gangliosides closely associate with single- or multiple-signal transducer molecules in the sphingolipid-enriched domains from neurones. Ganglioside GD3 has been shown to be associated with Src-family kinase Lyn and the neural cell adhesion molecule TAG-1 in rat brain (Kasahara et al. 1997, 2000). The association was strong enough to allow protein co-immunoprecipitation using anti-GD3 monoclonal antibody. A strong association between sphingolipid and protein components of the sphingolipid-enriched domains was also demonstrated by immunoprecipitation experiments using anti-c-Src, anti-Lyn and anti-Csk antibodies (Prinetti et al. 2000b). The immunoprecipitation experiments were carried out on SEMFs prepared from [1-3H]sphingosine metabolically labelled cells, under experimental conditions that preserved the integrity of the domain. Using this method, we obtained immunoprecipitates that contained a significant amount of radioactive sphingolipids. These data suggested that at least an annular lipid environment, containing different sphingolipid molecules, was strongly associated with each protein. The composition of these lipid domains indicated the presence of a complex organization of sphingolipid and protein molecules within the cell membrane. However, it was quantitatively different from that of the SEMF, suggesting the possible existence of different lipid membrane domains.

In the present paper, we analysed the composition of immunoprecipitates obtained using an anti-ganglioside antibody from a cell lysate prepared under experimental conditions that preserved the integrity of the domain, or from a SEMF. We used [1-3H]sphingosine metabolically labelled rat cerebellar granule cells differentiated in culture to allow the sensitive quantitative analysis of the sphingolipids associated with each fraction. Mouse anti-GD3 monoclonal antibody R24 was used, because of its availability and its wide use in many different experimental approaches, including the study of ganglioside/protein interactions within sphingolipid-enriched domains (Kasahara et al. 1997, 2000).

In agreement with previous data (Prinetti et al. 2000a, 2001), the SEMFs prepared from rat cerebellar granule cells were highly enriched in sphingolipids and in some functional proteins (Kasahara et al. 1997, 2000; Prinetti et al. 2000a,b, 2001). In fact, the SEMF contained about 60% of total cell sphingolipids. Among proteins, the amount of TAG-1, Lyn, Fyn and c-Src in this fraction was 32, 6, 43 and 39% of their total cellular content. Because the SEMF prepared from these cells only contained about 1.6% of total cell proteins (Prinetti et al. 2000a), SEMF enrichment in TAG-1, Lyn, Fyn and c-Src was remarkable. Figure 1 shows the patterns of these proteins in the SEMF obtained from rat cerebellar granule cells differentiated in culture.

Figure 1.

 Protein patterns in the sphingolipid-enriched membrane fraction (SEMF) prepared from rat cerebellar granule cells differentiated in culture. The same percentage of the total from the low-density SEMF (lane 1) and from the high-density fractions of the sucrose gradient (lane 2) were analysed by SDS–PAGE followed by western blotting detection using specific antibodies (lower part), as indicated at the top of each panel. The relative quantities of each protein in the fractions (upper part) were calculated by densitometry on the basis of the intensity of the blotting signal. The right-hand panel show the relative content of cellular protein in each fraction.

Figures 2 and 3 show the TLC patterns of radioactive lipids (lower part) and the immunoblotting with anti-TAG-1, anti-Lyn, anti-Fyn and antic-Src antibodies (upper part), associated with the immunoprecipitates obtained from the cell lysate (Fig. 2, lane 4) and from the detergent-resistant SEMF (Fig. 3, lane 4). The radioactive lipid patterns of these immunoprecipitates were very similar, and closely resembled that of the SEMF (Fig. 3, lane 1). In both cases, the immunoprecipitates were enriched in TAG-1, Lyn, Fyn and c-Src. Moreover, the immunoprecipitate obtained from the cell lysate did not contain phosphatidylethanolamine; this glycerophospholipid, clearly detectable in the total lipid extract from the cell lysate (Fig. 2, lane 1), as 8.5% of total lipid associated radioactivity, was found in the corresponding supernatant after separation of the immunoprecipitate by centrifugation (Fig. 2, lane 6).

Figure 2.

 Association of radioactive lipids (lower panel) and proteins (upper panel) with the immunoprecipitates obtained from rat cerebellar granule cell lysate. Lipids were extracted, separated by HPTLC and detected by digital autoradiography (30–500 dpm were applied on a 3-mm line; acquisition time, 48 h). Lane 1, total lipid pattern; lane 2, lipid pattern associated with protein G-coupled magnetic beads used to preclear the sample; lane 3, lipid pattern associated with negative control; lane 4, lipid pattern associated with the anti-GD3 monoclonal antibody R24 immunoprecipitate; lane 5, lipid pattern associated with the supernatant obtained after separation of negative control immunoprecipitate; lane 6, lipid pattern associated with the supernatant obtained after separation of anti-GD3 immunoprecipitate. PE, phosphatidylethanolamine. Sphingolipid nomenclature follows the IUPAC-IUBMB (1997, 1998) recommendations. Proteins were analysed by SDS–PAGE followed by western blotting detection using specific antibodies, as indicated at the left of each panel.

Figure 3.

 Association of radioactive lipids (lower panel) and proteins (upper panel) with the immunoprecipitates obtained from the detergent-resistant sphingolipid-enriched membrane fraction from rat cerebellar granule cell. Lipids were extracted, separated by HPTLC and detected by digital autoradiography (30–500 dpm were applied on a 3-mm line; acquisition time, 48 h). Lanes are numbered as in the legend of Fig. 2. Proteins were analysed by SDS–PAGE followed by western blotting detection using specific antibodies, as indicated at the left of each panel.

Data previously reported (Kasahara et al. 1997) suggested that the anti-GD3 antibody was able to immunoprecipitate Lyn together with GD3 ganglioside. The results reported in Figs 2 and 3 clearly indicate that the anti-GD3 is able to immunoprecipitate a complex lipid domain containing the same proteins and sphingolipids detected in the detergent-insoluble SEMF. Thus, the question is, whether the anti-GD3 antibody immunoprecipitate corresponds to a lipid membrane domain. Cholesterol and phosphatidylcholine, the two main lipids of the sphingolipid-enriched membrane domains (Prinetti et al. 2001), were found in the immunoprecipitates. These data were obtained by TLC separation of the total lipid extracts followed by colourimetric detection (Fig. 4). The scant amounts of the lipid extracts and the presence of several contaminants did not allow a precise quantification of cholesterol and phosphatidylcholine. Nevertheless, from the colour intensity of the separated cholesterol and phosphatidylcholine, compared with standard compounds, it is possible to derive that the quantities of both lipids associated with the immunoprecipitates are not too far from those in the SEMF. Therefore, the immunoprecipitates obtained by anti-GD3 antibody, containing cholesterol, phosphatidylcholine, sphingolipids, TAG-1, Lyn, Fyn and c-Src, have composition that resembles that of the sphingolipid-enriched membrane domain.

Figure 4.

 HPTLC analysis of the total lipid extract from the anti-GD3 immunoprecipitate. Experimental conditions were used allowing the analysis of cholesterol (solvent system: hexane/diethylether/acetic acid, 80 : 20 : 1 v/v, panel a), and phosphatidylcholine (solvent systems: chloroform/methanol, 9 : 1 v/v, followed by chloroform/methanol/acetone/acetic acid/water, 10 : 2 : 4 : 2:1 v/v, panel b) Lane 1, cholesterol standard (panel a), or phosphatidylcholine standard (panel b); lane 2, lipid extract from the negative control; lane 3, lipid extract from the anti-GD3 immunoprecipitate obtained from the sphingolipid-enriched membrane fraction.

A certain quantity of radioactivity was associated with the negative controls and with the protein G-coupled magnetic beads used to preclear samples. The quantity of radioactivity (Table 1) and the lipid pattern (Figs 2 and 3, lanes 2 and 3) associated with these samples were very similar, suggesting a non-specific interaction between sphingolipids and protein G-coupled magnetic beads. Moreover, the fact that these samples did not contain the above proteins, suggests that the non-specific interaction involves sphingolipids not belonging to the domains.

Table 1.   Distribution of the radioactivity after immunoprecipitation with anti-GD3 antibody in cell lysate and sphingolipid-enriched membrane fraction from [1-3H]sphingosine labelled cerebellar granule cells differentiated in culture
 TotalPreclearImmunoprecipitateSupernatant
  1. Data are expressed in dpm.

Cell lysate
Normal mouse IgG790 30092 30089 300571 300
Anti-GD3790 300110 100167 500501 300
Sphingolipid-enriched membrane fraction
Normal mouse IgG380 50014 50015 100305 000
Anti-GD3380 50021 100153 650200 200

The preparation of the sphingolipid-enriched membrane domains is based on the ability to preserve the integrity of the domains when cells are lysed in the presence of 1% Triton X-100 at 4°C (Brown and Rose 1992). In fact, under these experimental conditions, the detergent does mix with the components of the cell membrane, but not with the rigid lipid domain, which can be separated from the bulk cell components on the basis of its low density by gradient centrifugation. Figure 5 shows the TLC patterns of radioactive lipids (lower part) and the patterns of TAG-1, Lyn, Fyn and c-Src obtained by immunoblotting using specific antibodies (upper part), associated with the immunoprecipitates from cell lysates prepared under two experimental conditions different from those designed to preserve the lipid domain integrity. Under one condition, the cell lysate was warmed at 37°C for 20 min to allow the disaggregation of the sphingolipid-enriched domains before performing the immunoprecipitation (lane 5). Under the other, the immunoprecipitation was performed using RIPA buffer that, together with 1% Triton X-100, also contained 0.2% SDS and 0.5% sodium deoxycholate (lane 6). Under both conditions, TAG-1, Lyn, Fyn and c-Src were not detected in the immunoprecipitates. In addition to this, the quantity of radioactivity associated with the immunoprecipitates was lower (lane 6), or much lower (lane 5), than that associated with the immunoprecipitate obtained under conditions that preserve the domain integrity.

Figure 5.

 Association of radioactive lipids (lower panel) and proteins (upper panel) with the immunoprecipitates obtained from rat cerebellar granule cell lysate under different experimental conditions: in IP buffer containing 1% Triton X-100 at + 4° (lanes 1 and 4), in IP buffer containing 1% Triton X-100 after 20 min incubation at + 37°C (lanes 2 and 5) and in RIPA buffer (lanes 3 and 6). Lanes 1–3, lipid patterns associated with negative controls; lanes 4–6, lipid patterns associated with the anti-GD3 monoclonal antibody R24 immunoprecipitates. Lipids were extracted, separated by HPTLC and detected by digital autoradiography (30–500 dpm were applied on a 3-mm line; acquisition time, 48 h; PE, phosphatidylethanolamine). Sphingolipid nomenclature follows the IUPAC-IUBMB (1997, 1998) recommendations. Proteins were analysed by SDS–PAGE followed by western blotting detection using specific antibodies, as indicated at the left of each panel.

In conclusion, our results would suggest that the anti-GD3 antibody forms a multimolecular complex with the whole membrane lipid domain through the coupling with GD3 oligosaccharide, allowing the immunoseparation of the domain itself. Thus, a network of interactions between lipids (including GD3) and proteins is required to maintain the organization of the lipid domain. Previous papers (Hencke et al. 1996; Iwabuchi et al. 1998; Chigorno et al. 2000; Vyas et al. 2001) suggested that the detergent insoluble membrane fraction prepared by sucrose gradient centrifugation is a heterogeneous mixture of different lipid membrane domains. The use of different antibodies against protein or lipid components of the lipid membrane domain fraction, together with the development of very sensitive analytical procedures, should allow us to obtain new pieces of information about the properties of the different membrane lipid domains.

Acknowledgements

We thank Miss Claudia Giannotta, Miss Elena Ottico and Dr Maria Traka for their skilful technical assistance.

Ancillary