Complete removal of sphingolipids from the plasma membrane disrupts cell to substratum adhesion of mouse melanoma cells.

GM-95, a mutant cell line derived from mouse melanoma MEB-4 cells, is deficient in glycosphingolipids (GSLs) due to the lack of ceramide glucosyltransferase-1 activity (Ichikawa, S., Nakajo, N., Sakiyama, H., and Hirabayashi, Y. (1994) Proc. Natl. Acad. Sci. U. S. A. 91, 2703-2707). In this study, we examined the involvement of the complex sphingolipids in cell to substratum adhesion. Immunofluorescent and chemical analyses revealed that the complex sphingolipids were significantly concentrated in the detergent-insoluble substrate attachment matrix of both GM-95 and MEB-4 cells. In spite of the absence of GSLs, GM-95 cells retained the ability to adhere to extracellular matrix (ECM) proteins such as fibronectin, collagen, and laminin. When both GM-95 and MEB-4 cells were treated with neutral sphingomyelinase, GM-95 cells were rounded up and detached from all ECM proteins examined. In contrast, neither the morphology nor the adherence of MEB-4 cells was altered. Under this treatment, sphingomyelin (SM) became undetectable in both cells. A similar inhibition was observed upon pretreatment of cells with fumonisin B1 or ISP-1, both of which block the synthesis of ceramide, a common precursor of both GSLs and SM. Stable transfectants expressing GSLs, which were established by transfection of glucosyltransferase-1 cDNA into GM-95 cells, became resistant to neutral sphingomyelinase-mediated rounding up and detachment from ECM proteins. In conclusion, the complex sphingolipids play critical roles in cell to substratum adhesion, and the presence of either GSLs or SM is sufficient for the adhesion.

are ubiquitous components of the eukaryotic plasma membrane and comprise approximately 10% of the plasma membrane lipids. Complex sphingolipids have the hydrophobic residue, ceramide, which consists of sphingosine and fatty acids. Recently, ceramide and sphingosine, as the breakdown products from SM, are reported to be involved in intracellular signaling pathways for various cytokines. They have been recognized as bioactive compounds that modulate protein kinase or ion channel activities (1)(2)(3).
GSLs have been defined as tumor antigens and as regional and temporal markers in early embryogenesis (4,5). Several lines of evidence have suggested that GSLs are involved in cell growth, differentiation, and adhesion (5,6). However, in spite of a large number of studies done during the past few decades, the physiological functions of GSLs have still been elusive.
We have established a GSL-deficient mutant cell line, GM-95, from a B16 mouse melanoma cell line, MEB-4, and determined that the mutant is deficient in ceramide glucosyltransferase (UDP-glucose:N-acylsphingosine D-glucosyltransferase; GlcT-1) that catalyzes the first step of GSL synthesis (7). We have also found that GM-95 cells can proliferate in a GSL-free medium, clearly indicating that GSLs are not essential for cell growth (7). Recently, Hanada et al. (8) have established a temperature-sensitive mutant cell line (SPB-1 cells) that is deficient in serine palmitoyltransferase and is responsible for the initial step of ceramide synthesis. At a non-permissive temperature, SPB-1 ceased growing in a sphingolipid-free medium (8,9). Although detailed mechanisms are not yet clear, their studies demonstrated that sphingolipids play critical roles in cell growth.
In spite of the physiological significance of sphingolipids, however, the biological function of membrane GSLs and SM is still unknown. GSLs have been implicated in cell to substratum adhesion, especially fibronectin (FN)-mediated adhesion (10 -12). GM-95 cells are expected to be an ideal tool not only to investigate the physiological function of GSLs but also to assess the biological significance of SM. Very recently, we have cloned a cDNA encoding GlcT-1 by expression cloning and have established stable transfectants expressing G M3 ganglioside on the surface of GM-95 cells (13). In the current study, we examined the role of complex sphingolipids including GSLs and sphingomyelin for cell to substratum adhesion by using the mutant cells and the GlcT-1 cDNA transfectants.

EXPERIMENTAL PROCEDURES
Materials-E-RDF medium was purchased from Kyokuto Pharmaceutical Corp. (Tokyo). The medium is a complete serum-free medium that contains insulin and transferrin (7). FN (human plasma), type IV collagen (bovine placenta), and laminin (mouse Engelbreth-Holm-Swarm sarcoma) were purchased from Kohken (Tokyo). Type I collagen (porcine skin) was obtained from Nitta Gelatin Co., Ltd. (Tokyo). Highly purified recombinant neutral sphingomyelinase (Bacillus cereus) and phospholipase C (B. cereus) were purchased from Higeta Shouyu Co., Ltd. (Tokyo) and Boehringer Mannheim, respectively. Fumonisin B1 was purchased from Sigma. ISP-1 was generously donated by Dr. T. Kawasaki, Department of Biological Chemistry, Kyoto University, Japan. Mouse anti-G M3 ganglioside monoclonal antibody (mAb) M2590 was purchased from Meiji Seika Kaisha, Ltd. (Tokyo). Mouse antisphingomyelin mAb VJ41 was donated by Dr. K. Ono, Pharma Research Laboratory, Hoechst Japan Limited. All other chemicals were of the highest purity available.
Cell to Substratum Adhesion Assay-The adhesion assay was performed as described previously (15). Briefly, plates (96-well) were coated with FN (5 g/ml), type I collagen (20 g/ml), type IV collagen (20 g/ml), laminin (20 g/ml), or BSA (20 g/ml) at 37°C for 16 h and then blocked with BSA (20 g/ml) at 37°C for 1 h. Cells (3 ϫ 10 4 ) were allowed to attach to the coated plates at 37°C for 24 h in a serum-free E-RDF medium and then incubated with N-SMase (1 unit/ml) at 37°C for the indicated times. The plates were inverted and centrifuged at 500 rpm for 3 min to remove non-adherent cells. Cells were fixed with methanol at room temperature for 10 min and stained with 1% (w/v) crystal violet at room temperature for 10 min. After washing five times with Tris-buffered saline (25 mM Tris, pH 7.4, 150 mM NaCl), the cells were lysed with 1% (w/v) sodium deoxycholate. The absorbance of each well was measured at 595 nm (with reference to absorbance at 405 nm) by a micro-plate reader. The cell number and the absorbance showed linear correlation between 1 and 8 ϫ 10 4 .
To examine the effect of sphingolipid inhibitors on cell adhesion, the cells (1 ϫ 10 4 ) were seeded onto a 96-well plate in serum-free E-RDF medium and incubated at 37°C for 72 h in the presence of indicated concentrations of inhibitors. Cells were then subjected to cell adhesion assay as described above.
To analyze lipids in substrate attachment matrix (SAM), the cells (2 ϫ 10 7 cells) were detached with PBS containing 3.8 mM EDTA (16). Lipids of the substrates were extracted sequentially with methanol and chloroform/methanol (1:2, v/v). Lipids extracted from SAM of 4 -5 ϫ 10 6 cells were applied per lane on TLC.
Establishment of Stable Transfectants with GlcT-1 cDNA-We have isolated GlcT-1 cDNA by expression cloning and designated it as pCG1 (13). The cDNA encoding GlcT-1 was cloned into a mammalian expression vector, pcDNA I (Invitrogen). Stable transfectants were established by cotransfection of pCG1 and pSV2neo (17) into GM-95 cells. Negative control clones were made by transfection with pcDNAI and pSV2neo. After G418 (0.8 mg/ml) selection, five G M3 -positive and three G M3 -negative independent clones were chosen and subjected to the cell adhesion assay and the lipid analysis.
Immunocytochemical Analysis-Cells were grown on glass coverslips in minimum Eagle's medium supplemented with 10% fetal calf serum, 100 units/ml penicillin, and 100 g/ml streptomycin at 37°C for 3 days. Cells were removed with PBS containing 3.8 mM EDTA as described previously (16), and substrates remaining on the coverslips were fixed with 4% paraformaldehyde, 0.1% glutaraldehyde, 0.2% picric acid in PBS, pH 7.2, at room temperature for 10 min. After washing three times with PBS, the substrates were incubated with primary antibodies (M2590 and VJ41) at 37°C for 2 h. The substrates were washed three times with PBS containing 0.25% BSA and 0.5% normal goat serum and then reacted with fluorescein isothiocyanate-conjugated goat antimouse IgM antibody at 37°C for 2 h.

RESULTS
Characterization of Lipids in GM-95, Glycosphingolipid-deficient Cell Line-We first examined lipid compositions of GM-95 and MEB-4 cells (Table I). MEB-4 cells had two major GSLs (glucosylceramide and G M3 ganglioside) and SM, whereas GM-95 cells had no GSLs as previously reported (7). The cells had SM as a sole complex sphingolipid. Interestingly, the SM content of GM-95 cells increased over 1.5-fold when compared with that of MEB-4 cells. Consequently, the total amounts of complex sphingolipids in MEB-4 and GM-95 cells were almost equal. Among other phospholipids, the phosphatidylcholine content of GM-95 cells decreased by less than 17%, compared with that of MEB-4 cells. Previous studies have shown that SM is mainly synthesized by phosphatidylcholine: ceramide cholinephosphotransferase that catalyzes a base-exchange reaction between ceramide and phosphatidylcholine (18). The reduction of the phosphatidylcholine content in GM-95 cells possibly resulted from consumption of the lipid for SM synthesis by the exchange enzyme. The content of other lipids such as cholesterol and phosphatidylethanolamine was not altered in either cell line.
Requirement of Sphingolipids for Cell to Substratum Adhesion-Participation of GSLs in cell to substratum adhesion is still controversial. There are many studies suggesting that GSLs play an important role in cell to substratum adhesion (5,6). Several others, however, indicate that sugar chains of GSLs do not affect the adhesion (10,19). To answer this question, we compared the adhesion ability of MEB-4 and GM-95 cells with several ECM proteins and examined the effects of N-SMase on the attachment (Figs. 1 and 2). GM-95 cells adhered to all ECM proteins examined (Fig. 1). Attachment of GM-95 cells to FN, however, was 60 -70% of that of MEB-4 cells. The surface expression of ␣5 integrin, an FN receptor, was at the same level in both GM-95 and MEB-4 cells when examined by fluorescence-activated cell sorter analysis with anti-␣5 integrin mAb (data not shown). These results support the previous reports that GSLs modulate cell to substratum adhesion (20,21). N-SMase treatment of GM-95 cells caused the rounding up of GM-95 cells ( Fig. 2A) and reduced the adhesion to ECM proteins examined. The treatment, however, affected neither the morphology nor the adhesion of MEB-4 cells (Fig. 2, A and  B).The effects were dependent on incubation time and concentration of N-SMase (Fig. 2, C and D). N-SMase treatment did not significantly affect the viability of either cell line (data not shown). Furthermore, the adhesion of GM-95 cells was restored by removal of N-SMase, meaning that any effect was reversible (Fig. 3). In addition, phospholipase C treatment had no apparent influence on cell adhesion in either cell line (data not shown). Therefore, the decreased adhesion resulted from removal of SM by N-SMase treatment.
We further investigated the alteration of the lipid composition by N-SMase treatment. After N-SMase treatment, SM was neither chemically detectable (Fig. 4) nor immunostained with VJ41, an anti-SM mAb (data not shown), indicating loss of all Complex Sphingolipids in Cell to Substratum Adhesion complex sphingolipids from GM-95 cells, whereas the GSL content of MEB-4 cells did not change by this enzyme treatment. The treatment did not significantly alter the relative content of the other phospholipids. Effect of Sphingolipid Inhibitors on Cell to Substratum Adhesion-When cells were stimulated with various kinds of agonists, sphingolipid-metabolic intermediates such as ceramide, sphingosine, and sphingosine 1-phosphate were shown to be produced from SM by cellular SMase. The metabolites exert diverse biological responses (1-3). Among them, sphingosine was shown to stimulate cell to substratum adhesion (22). Therefore, there is a possibility that these intermediates produced by N-SMase are involved in the loss of cell to substratum adhesion. To exclude this possibility, we used two different metabolic inhibitors for the synthesis of ceramide, which is a common precursor of all complex sphingolipids, including GSLs and SM. Fumonisin B1 inhibits dihydrosphingosine acyltransferase responsible for dihydroceramide synthesis (23). ISP-1 inhibits serine palmitoyltransferase by which 3-ketosphinganine is formed (24). Thus, these inhibitors were expected to be useful reagents to clarify involvement of the intermediates in cell to substratum adhesion. In particular, ISP-1 could exclude any possibility that metabolic intermediates affect adhesion because the inhibitor blocks the initial step of sphingolipid synthesis. By treatment with ISP-1 (5 M), the adhesion ability of MEB-4 and GM-95 cells was reduced to 70 and 40% of that of the control, respectively (Fig. 5), whereas fumonisin B1 (100 M) inhibited the adhesion of GM-95 cells significantly (60% of that of the control) but not that of MEB-4 cells. To confirm the effect of these inhibitors on sphingolipid biosynthesis, we examined the complex sphingolipid contents of both cells after the inhibitor treatments (Table II). ISP-1 and fumonisin B1 reduced the total amount of complex sphingolipids in both cells to less than 20 and 30% of that of the control, respectively. In contrast to SM, the content of GSLs, particularly glucosylceramide, in MEB-4 cells was drastically decreased after ISP-1 and fumonisin B1 treatments (1.6 and 8.2% of the control, respectively). Although the complex sphingolipid contents of MEB-4 and GM-95 cells were reduced similarly by the treatment, the effect on the adhesion was more dominant in GM-95 cells than that in MEB-4 cells. Because all GSLs could not be eliminated from MEB-4 cells after the treatment, the remaining GSLs may mediate cell to substratum adhesion more effectively than SM. This finding suggests that sugar chains bound to ceramide may contribute to the cell to substratum adhesion more potently than phosphocholine bound to it. From the above observations, we concluded that the reduction of all sphingolipids, but not the accumulation of metabolic intermediates, caused the loss of cell to substratum adhesion.
Prevention of N-SMase Effect by Transfection of GlcT-1 cDNA into GM-95 Cells-We have isolated a cDNA encoding GlcT-1 and established stable transfectants by introducing the cDNA into GM-95 cells (13). The transfectants (CG1 through CG5) expressed GSLs at different levels (Table III). Irrespective of the expression level of GSLs, these cells became resistant to the effects of N-SMase on the cell adhesion to substratum to an extent similar to that of MEB-4 cells (Fig. 6). By N-SMase treatment, SM in all transfectants was undetectable (data not shown). These findings supported a conclusion that the presence of either GSLs or SM is sufficient for the cell to substratum adhesion.
Distribution of Complex Sphingolipids in the SAM-GSLs were shown to be present in SAM (11,12,16). Here, we found that SM was also enriched in the focal contact of both MEB-4 and GM-95 cells (Table IV). In both cells, SM content of SAM was more than double that of whole cells. However, the GSL content of MEB-4 cells in whole cells and SAM showed no difference. We also stained complex sphingolipids in the SAM using M2590 and VJ41 mAbs. Fig. 7 shows the existence of G M3 ganglioside and SM in the SAM. In MEB-4 cells, G M3 ganglioside was clearly stained in SAM as described previously (16). However, sphingomyelin was unexpectedly not detectable by the VJ41 mAb despite evidence of its existence. Although VJ41 mAb strongly stained SM in the SAM of GM-95 cells, it also became undetectable after the transfection of GlcT-1 cDNA (data not shown). It is likely that once GSLs were expressed on   Complex Sphingolipids in Cell to Substratum Adhesion the plasma membrane, SM became inaccessible to the antibody. These observations indicated that complex sphingolipids, particularly SM, in the SAM of both cells were much more concentrated than other lipids.

DISCUSSION
In the present study, we demonstrated that when cells lose all complex sphingolipids, they lose the ability to adhere to ECM proteins. Since GM-95 cells possessed SM as the sole complex sphingolipid, the mutant allowed us to test the role of GSLs and SM in cell to substratum adhesion. N-SMase treatment caused detachment of GM-95 cells from ECM proteins, but the same treatment affected neither the morphology nor the attachment of MEB-4 cells that contained GSLs. Transfection of GlcT-1 cDNA into GM-95 cells restores the expression of GSLs, and the resulting transfectants became resistant to N-SMase-mediated cell detachment. These results suggest that either GSLs or SM is required for cell to substratum adhesion. We think that metabolic intermediates such as sphingosine and ceramide are not responsible for the phenomenon, based on the following observations: 1) Although the intermediates were produced after N-SMase treatment, MEB-4 cells still attached to ECM proteins. Under this condition, GSLs stayed at a normal level in the cells, whereas complete removal of SM by N-SMase treatment caused detachment of GM-95 cells from ECM proteins. 2) The adhesion of both cells was decreased by the treatment of sphingolipid inhibitors that blocked biosyn-thesis of all sphingolipids, including their metabolic intermediates such as sphingosine and ceramide. 3) Neither exogenously added ceramide nor sphingosine affected the adhesion of MEB-4 and GM-95 cells (data not shown). Involvement of complex sphingolipids in cell to substratum adhesion was further supported by the observations that GSLs and SM were immunostained in SAM of MEB-4 and GM-95 cells, respectively, and that the relative amount of SM in the SAM was much higher than that in the whole cells of both cell lines. Although M2590 mAb, the anti-G M3 antibody, preferentially stained the focal contacts (Fig. 7), the biochemical analysis in Table IV showed only modest enrichment of this antigen in the matrix fraction. This is not so surprising since there are several environmental factors involved in the binding of M2590 to a melanoma antigen, G M3 ganglioside, as previously discussed (25). In the present study, we had another typical example in which a secondary factor participated in the generation of the antigenic structure of a membrane lipid. In the case of sphingomyelin antigen recognized by VJ41 mAb, once GSLs were expressed on the plasma membrane, the reactivity of the antibody disappeared in spite of its existence (Fig. 7). GSLs, most probably G M3 gangliosides, may control the degree of accessibility of VJ41 mAb to the antigen on the cell surface.
Several studies showed that GSLs were enriched in the SAM (11,12,26). In BHK cells, GSL content in the SAM is over 2-fold higher than that in whole cells. Although GSLs in MEB-4 cells were present in the SAM, they were not significantly enriched. This may be due to the difference of cell type or species. There is also considerable evidence demonstrating that some GSLs are concentrated at the focal contact and interact with integrins (11,12,16,20,(27)(28)(29)(30). Integrins are also concentrated at the focal contact or focal adhesion (31). Both MEB-4 and GM-95 cells expressed ␣5 integrin at the same level, and treatment of the cells with anti-␣5 integrin mAb inhibited the adherence to FN (data not shown). Thus, the adhesion of the cells to ECM proteins is postulated to be mediated by complex sphingolipids and integrins at the focal contact. The absence of complex sphingolipids may disrupt the adhesive function of integrins.
We have shown that the presence of either GSLs or SM is sufficient for maintenance of cell to substratum adhesion, irrespective of the polar head groups (sugar chains (GSLs) or phosphocholine (SM)). Thus, SM and GSLs are thought to play similar roles in cell to substratum adhesion, and the polar head group of complex sphingolipids may not have strict specificity in the function. It is likely that head groups are necessary to retain sphingolipids on the outer leaflet of the plasma mem-  brane. In fact, a fluorescent derivative of SM incorporated into the plasma membrane was quickly internalized after N-SMase treatment (data not shown). We have also examined the lipid composition of GM-95 and MEB-4 cells in detail. The SM content of GM-95 cells was significantly higher than that of MEB-4 cells. Consequently, the total amount of complex sphingolipids was identical between MEB-4 and GM-95 cells, suggesting that the cellular content of complex sphingolipids is maintained at a certain level. Presumably, factors responsible for ceramide synthesis regulate the total content of complex sphingolipids. In conclusion, we have proven that complex sphingolipids are essential for cell to substratum adhesion. Since cell to substratum adhesion is important for cell growth and differentiation (32,33), it is possible that complex sphingolipids also play important roles in such cellular processes.