Interaction of GM2 activator protein with glycosphingolipids.

GM2 activator protein is a protein cofactor that has been shown to stimulate the enzymatic hydrolysis of both GalNAc and NeuAc from GM2 (Wu, Y. Y., Lockyer, J. M., Sugiyama, E., Pavlova, N.V., Li, Y.-T., and Li, S.-C. (1994) J. Biol. Chem. 269, 16276-16283). To understand the mechanism by which GM2 activator stimulates the hydrolysis of GM2, we examined the interaction of this activator protein with GM2 as well as with other glycosphingolipids by TLC overlay and Sephacryl S-200 gel filtration. The TLC overlay analysis unveiled the binding specificity of GM2 activator, which was not previously revealed. Under the conditions optimal for the activator protein to stimulate the hydrolysis of GM2 by beta-hexosaminidase A, GM2 activator was found to bind avidly to acidic glycosphingolipids, including gangliosides and sulfated glycosphingolipids, but not to neutral glycosphingolipids. The gangliosides devoid of sialic acids, such as asialo-GM1 and asialo-GM2, and the GM2 derivatives whose carboxyl function in the NeuAc had been modified by methyl esterification or reduction, were only very weakly bound to GM2 activator. These results indicate that the negatively charged sugar residue or sulfate group in gangliosides is one of the important sites recognized by GM2 activator. For comparison, we also studied in parallel the complex formation between glycosphingolipids and saposin B, a separate activator protein with broad specificity to stimulate the hydrolysis of various glycosphingolipids. We found that saposin B bound to neutral glycosphingolipids and gangliosides equally well, and there was an exceptionally strong binding to sulfatide. In contrast to previous reports, we found that GM2 activator formed complexes with GM2 and other gangliosides in different proportions depending on the ratio between the activator protein and the ganglioside in the incubation mixture prior to gel filtration. We were not able to detect the specific binding of GM2 activator to GM2 when GM2 was mixed with GM1 or GM3. Thus, the specificity or the mode of action of GM2 activator cannot be simply explained by its interaction with glycosphingolipids based on complex formation. The binding of GM2 activator to a wide variety of negatively charged glycosphingolipids may indicate that this activator protein has functions other than assisting the enzymatic hydrolysis of GM2.

In higher animals, the sugar chains of glycosphingolipids are catabolized by the sequential action of lysosomal exoglycosidases (1). It has been shown that, in addition to ␤-hexosamini-dase A, the conversion of G M2 1 into G M3 requires the assistance of G M2 activator, a low molecular weight protein cofactor (2)(3)(4). The physiological significance of G M2 activator has been demonstrated by the fact that the congenital defect of this activator protein leads to cerebral accumulation of G M2 in type AB Tay-Sachs disease (5,6).
Human G M2 activator has been isolated from kidney (4), brain (6), and liver (7). This activator has been shown to be very specific in stimulating the hydrolysis of GalNAc from G M2 by ␤-hexosaminidase A (1,4,7). This activator protein was also shown to assist the hydrolysis of NeuAc from G M2 by clostridial sialidase (8) and to recognize the branched trisaccharide (G M2epitope) in G M2 (9). This activator, however, is not required for the hydrolysis of water-soluble synthetic substrates such as 4-methylumbelliferyl-␤-GlcNAc or p-nitrophenyl-␤-GlcNAc by ␤-hexosaminidase A. The mode of action of G M2 activator is still not well understood. Through the studies of complex formation between G M2 activator and glycosphingolipids using electrophoresis, isoelectric focusing, and ultracentrifugation (4, 10), Conzelmann and Sandhoff (4) postulated that the action of G M2 activator is to extract a single G M2 molecule from its micelles to form a water-soluble protein-lipid complex (1:1 ratio), which serves as the true substrate for ␤-hexosaminidase A. This hypothesis, however, is not supported by two simple facts: (a) The water-soluble tetrasaccharide derived from G M2 cannot be hydrolyzed by ␤-hexosaminidase A in the presence or absence of the activator (8) and (b) saposin B, another activator protein whose action is to solubilize glycosphingolipids, does not stimulate the hydrolysis of G M2 by ␤-hexosaminidase A.
The results of previous studies on the interaction between glycosphingolipids and the activator proteins isolated from human tissues might have been complicated by the possible presence of contaminated proteins. Recently, we have cloned the cDNA encoding human G M2 activator (11) and also expressed the cDNA in Escherichia coli (8). The availability of pure recombinant human G M2 activator in large quantities made the re-examination of the interactions between G M2 activator and glycosphingolipids possible. To understand the mode of action of G M2 activator, we have studied the interaction of G M2 activator with various glycosphingolipids by TLC overlay and Sephacryl S-200 gel filtration. For comparison, we have also studied in parallel the interaction of glycosphingolipids with saposin B, a nonspecific activator protein that has been reported to stimulate the enzymatic hydrolysis of a wide variety of glycosphingolipids (12). We found that in aqueous medium, such as gel filtration, one molecule of G M2 activator was able to associate with multiple molecules of gangliosides. By TLC overlay, G M2 activator was found to bind to various negatively charged glycosphingolipids without showing preference to any particular sugar chain.
Radiolabeling of G M2 Activator and Saposin B-14 C-Labeled G M2 activator and saposin B were prepared by reductive methylation of amino groups with 14 C-labeled formaldehyde (23,24). Briefly, 300 g of G M2 activator or saposin B were dissolved in 85 l of 0.2 M phosphate buffer, pH 7.0. To this solution, 88.4 g of dimethylamine borane complex, which had been dissolved in 10 l of methanol, was added. After addition of 5 l (2.2 mol) of aqueous [ 14 C]formaldehyde, the mixture was left at room temperature for 6 h. Then, the resulting 14 C-labeled protein was separated from the reagents by gel filtration on a Bio Gel P-6 column (0.9 ϫ 10 cm) using water as an eluant, followed by dialysis against 10 mM ammonium acetate buffer, pH 6.8, and lyophilized. The stimulatory activities of the 14 C-labeled G M2 activator and saposin B on the hydrolyses of G M2 by ␤-hexosaminidase A and G M1 by ␤-galactosidase, respectively, were confirmed by the methods described previously (8,12).
TLC Overlay-The three buffer solutions used for studying the interactions between the activator proteins and glycosphingolipids on TLC plates were (a) 25 mM ammonium acetate buffer, pH 4.0, a low ionic strength acidic buffer providing the optimal condition for assaying the hydrolysis of G M2 by ␤-hexosaminidase A; (b) 25 mM ammonium acetate buffer, pH 6.8, a low ionic strength neutral buffer; and (c) 250 mM ammonium acetate buffer, pH 4.0, a high ionic strength acidic buffer. Each glycosphingolipid sample (10 -15 nmol) in chloroform: methanol (2/1 v/v) was first applied onto a Polygram SIL G TLC plate, and the plate was developed with chloroform:methanol:water (60/35/8, v/v/v). The dried plate was then immersed and kept at 37°C for 30 min in one of the above mentioned buffer solutions, which contained 1% each of polyvinylpyrrolidone and bovine serum albumin. The plate was then incubated in 5 ml of the same buffer solution containing 50 g of the 14 C-labeled activator protein (250,000 cpm) and 3% polyvinylpyrrolidone at 37°C for 1 h and washed three times with the buffer solution containing 0.05% Tween 20 and then air dried. Finally, the protein-lipid complexes were detected by placing the TLC plate onto an x-ray film to obtain a radioautogram. After obtaining the radioautogram, the same TLC plate was sprayed with diphenylamine reagent (25) and heated at 110 -120°C for 15-20 min to reveal the glycosphingolipids on the plate.
Gel Filtration Chromatography-For studying the complex formation between G M2 activator and the micellar form of glycosphingolipids using Sephacryl S-200 gel filtration, 25 mM ammonium acetate buffer, pH 4.0, was used as the incubation buffer and also to equilibrate and elute the column. G M2 activator (25 g, 1.34 nmol) in 100 l of the buffer solution was mixed with a given amount of a ganglioside or oligo-G M2 and incubated at 37°C for 30 min. The entire mixture was subsequently applied onto a Sephacryl S-200 column (0.6 ϫ 30 cm) connected to an HPLC system (Waters 600E, Millipore). The column was then eluted with the same buffer at a flow rate of 0.25 ml/min, and the effluent was monitored by the absorbance at 280 nm (Waters 490E UV-VIS detector). Fractions of 0.5 ml (2 min) were collected through the entire run, and each fraction was analyzed for the content of the activator protein and the glycosphingolipid.
Hydrolysis of G M2 -The fractions that contained the protein-lipid complex eluted from the Sephacryl S-200 column were incubated with 0.5 units of ␤-hexosaminidase A at 37°C for 3 h. Each incubated fraction was evaporated to dryness, dissolved in 20 l of chloroform: methanol (2/1, v/v), and analyzed by Silica gel 60 TLC plate using chloroform:methanol:water (60/35/8, v/v/v) as the developing solvent. Gangliosides were visualized by spraying the plate with diphenylamine reagent (25) followed by heating at 110 -120°C for 15-20 min.
Analytical Methods-When the activator protein was incubated with only [ 3 H]G M1 or [ 3 H]G M2 , the amount of the [ 3 H]G M1 or [ 3 H]G M2 in the protein-lipid complex was determined as follows: a 50-l aliquot of each fraction obtained from the Sephacryl S-200 column was mixed with 5 ml of Universol, and the radioactivity was measured by a Tri-Carb model 1600 CA liquid scintillation counter (Packard Instrument Co., IL). When the activator protein was incubated with G M3 , which was not radiolabeled, the amounts of G M3 in the protein-lipid complexes were determined by TLC analysis using the resorcinol spray (26) (27) and then individually scraped off the plate and mixed with Universol; the radioactivity was then measured by a scintillation counter. When the activator protein was incubated with both G M2 and G M3 , the amounts of G M2 and G M3 in the protein-lipid complex were determined as follows: an aliquot of each fraction was evaporated to dryness, redissolved in 20 l of chloroform:methanol (2/1, v/v), and applied onto a TLC plate. The plate was developed with the solvent system as described above for separating G M1 and G M2 , and the gangliosides were visualized with the diphenylamine reagent (25). The amounts of G M2 and G M3 were quantitated by scanning the TLC plate with a Scan Jet IICX and analyzed by NIH Image 1.55.
Determination of Protein-Protein was determined by the method of Lowry et al. (28) using bovine serum albumin as a standard.

RESULTS
Interaction of G M2 Activator with Glycosphingolipids on a TLC Plate-The interactions between G M2 activator and glycosphingolipids were examined by TLC overlay on which glycosphingolipids were associated with silica gel. Fig. 1A shows the representative common acidic and neutral glycosphingolipids on the plate that were stained by the diphenylamine reagent (25). While the same amount (15 nmol) of each glycosphingolipid was applied on the plate, GlcCer and GalCer showed weaker staining than G M1 , since the color intensity produced by the diphenylamine reagent depends on the sugar content of the glycosphingolipids. The same TLC plate prior to the chemical staining was overlaid with the radiolabeled G M2 activator as described under "Experimental Procedures," and the results are shown in Fig. 1B. The conditions for the overlay were first chosen to use the low ionic strength acidic buffer (25 mM ammonium acetate buffer, pH 4.0), which is the optimal condition for G M2 activator to stimulate the hydrolysis of G M2 by ␤-hexosaminidase A. Under this condition, G M2 activator protein binds avidly to gangliosides G M1 , G M2 , and G M3 (lanes 1, 2, and 3, respectively) but very weakly to the neutral glycosphingolipids, LacCer, GalCer, and GlcCer (lanes 4, 5, and 6, respectively). The bindings of G M2 activator to 14 other glycosphingolipids were further examined under the same conditions. As summarized in Table I, G M2 activator binds to several other acidic glycolipids such as G M4 , the synthesized PE-G M2 (8) which contains the oligosaccharide of G M2 linked to phosphati-dylethanolamine instead of ceramide, NeuGc-G M1 , KDN-G M3 and two chemically synthesized gangliosides, IV 6 KDNLcOse 4 Cer and IV 6 KDNLnOse 4 Cer, whose KDN residues are linked through ␣236Gal. These results indicate that the interactions between G M2 activator and glycosphingolipids require the presence of an acidic moiety on the glycosphingolipid, and the binding is not significantly affected by the sugar chain backbones, the position (␣233Gal versus ␣236Gal), and the nature of sialic acid (NeuAc versus NeuGc or KDN) in the glycosphingolipids. Furthermore, G M2 activator also binds to the sulfated glycosphingolipids, such as sulfatide and S M3 , but not to the asialogangliosides, such as G A1 and G A2 . These results strongly suggest that the recognition sites on the glycosphingolipids for the binding by G M2 activator are the anionic residues. These observations were further supported by studying the bindings between G M2 activator and the two chemically modified G M2 . As shown in Fig. 2, conversion of the carboxylic function of NeuAc in G M2 to a methyl ester (Me-G M2 ) (Fig. 2, lane 3) or to an alcohol (HO-G M2 ) (Fig. 2, lane 4) abolishes the ability of the two modified G M2 derivatives to interact with G M2 activator. The results of this binding study explain our previous observation that Me-G M2 and HO-G M2 were not hydrolyzed by ␤-hexosaminidase A in the presence of G M2 activator but could be hydrolyzed in the presence of sodium taurodeoxycholate (16). Fig. 2 also shows that sulfatide (lane 1) is very weakly stained by diphenylamine (25) due to its highly acidic sulfate residue. However, this acidic glycosphingolipid binds strongly to G M2 activator.
The extent of the bindings of G M2 activator to the glycosphingolipids was significantly reduced by raising the pH and the ionic strength of the buffer (Fig. 3). When the binding assay was carried out in a low ionic strength neutral pH buffer (pH 6.8), almost no bindngs between G M2 activator and glycosphingolipids were detected (Fig. 3, A and AЈ). Even sulfatide, which usually binds strongly to G M2 activator, was only very weakly bound to the activator protein under the neutral pH (lane 5). The binding detected in a high ionic strength acidic buffer (250 mM ammonium acetate, pH 4.0) was also considerably reduced (Fig. 3BЈ). G A2 (lane 6) showed no bindings with G M2 activator under all conditions tested. These results corroborated our previous observation that the conversion of GA 2 to LacCer was not effectively stimulated by G M2 activator (8), and the hydrolysis of the GalNAc from G M2 was greatly inhibited by the high ionic strength of the buffer solution; however, no such effect was observed for the hydrolysis of the GalNAc from GA 2 (asialo-G M2 ).
Interaction between Saposin B and Glycosphingolipids-Saposin B is a nonspecific activator protein that stimulates the enzymatic hydrolysis of a number of glycosphingolipids catalyzed by different glycosidases (12). This activator protein was reported to bind glycosphingolipids to form lipid-protein complexes (29,30). Therefore, the interactions between saposin B FIG. 2. Effect of the modification of carboxyl group of NeuAc in G M2 on the interaction with G M2 activator. A, the TLC plate with the indicated glycosphingolipids (10 nmol) was stained with diphenylamine reagent. B, the radioautogram of the same TLC plate that was overlaid with the 3 H-labeled G M2 activator prior to the chemical staining as shown in A. Other conditions for this experiment were identical to that used for Fig. 1. A and B: lanes 1, sulfatide; lanes 2, G A2 ; lanes 3   and glycosphingolipids were also examined in the same manner for comparison. As shown in Fig. 4, saposin B was found to bind not only to gangliosides and sulfatide but also to G A2 and LacCer. Compared with G M2 activator, saposin B bound to glycosphingolipids better at the neutral pH (pH 6.8) (Fig. 4BЈ), and the general behavior of binding was not greatly affected by the acidic pH (Fig. 4AЈ) or the high ionic strength of the buffer solution (Fig. 4CЈ).
Interaction of G M2 Activator with Gangliosides in Micellar Forms-Since the results of the TLC overlay experiment showed the preferential binding of G M2 activator to the anionic glycosphingolipids, we subsequently examined the interactions between the G M2 activator and the the gangliosides in aqueous medium using Sephacryl S-200 gel filtration to separate the protein-lipid complexes. Sephacryl S-200 column offers a special advantage for this analysis because this column adsorbs the free gangliosides but not the protein-lipid complexes. This enabled us to isolate and analyze the content in the complexes. When G M2 activator was applied alone to the column, the protein was not adsorbed and eluted from the column at the retention time of 28 min (Fig. 5A), whereas applying [ 3 H]G M2 alone, the ganglioside was retained by the Sephacryl S-200 gel. When an incubation mixture containing [ 3 H]G M2 and G M2 activator in a molar ratio of 1:1 or 50:1 was applied to the column, a peak containing both G M2 activator and [ 3 H]G M2 was eluted (Fig. 5, B and C). This peak was confirmed to be the proteinlipid complex by two separate analyses: (a) rechromatography of this complex did not result in the separation of the activator protein from [ 3 H]G M2 and (b) incubation of the complex with ␤-hexosaminidase A resulted the conversion of [ 3 H]G M2 into [ 3 H]G M3 (Fig. 6). In these experiments, the recoveries of the activator protein and the gangliosides were determined to be in the range of 57-72% and 69 -87%, respectively. As shown in Fig. 5, B and C, the complex derived from the incubation mixture that contained [ 3 H]G M2 and G M2 activator in a molar ratio of 50:1 (Fig. 5C) had a slightly shorter retention time and a broader peak area than that derived from the mixture that contained [ 3 H]G M2 and G M2 activator in an equimolar ratio (Fig. 5B). Similar chromatographic profiles were obtained when G M2 activator was incubated with either G M1 or G M3 . FIG. 3. Effect of pH (A and A) and ionic strength (B and B) of the buffer solutions on the complex formation between glycosphingolipids and G M2 activator. A and B, the TLC plates with the indicated glycosphingolipids (10 nmol) were stained with diphenylamine reagent. AЈ and BЈ, the radioautograms of the TLC plates that were overlaid with the 3 H-labeled G M2 activator prior to the chemical staining as shown in A and B, respectively. The buffer solutions used for TLC overlay were 25 mM ammonium acetate buffer, pH 6.8 (AЈ), and 250 mM ammonium acetate buffer, pH 4.0 (BЈ). Other conditions for this experiment were identical to that used for Fig. 1. Lane 1, G M1 ; lanes 2, G M2 ; lanes 3, G M3 ; lanes 4, LacCer; lanes 5, sulfatide; lanes 6, GA 2 .

FIG. 4. Detection of the complex formation between saposin B and glycosphingolipids by TLC overlay.
A-C, the TLC plates with the indicated glycosphingolipids (10 nmol) were stained with diphenylamine reagent. AЈ-CЈ, the radioautograms of the same TLC plates that were overlaid with the 3 H-labeled saposin B prior to the chemical staining as shown in A-C, respectively. The buffer solutions used for TLC overlay were 25 mM ammonium acetate buffer, pH 4.0 (AЈ), 25 mM ammonium acetate buffer, pH 6.8 (BЈ), and 250 mM ammonium acetate buffer, pH 4.0 (CЈ). Other conditions used were identical to that used for Fig. 1. In all panels, lanes 1, G M1 ; lanes 2, G M2 ; lanes 3 However, no complex formation was detected when G M2 activator was incubated with oligo-G M2 (data not shown) indicating that, in addition to the negative charge, the lipid moiety of the glycolipid is also essential for binding.
The complexes formed between G M2 activator and the different molar ratios of gangliosides were individually isolated from the Sephacryl S-200 column and analyzed for the ratio between the ganglioside and the activator protein. As shown in Table II, when G M2 activator was incubated with an equimolar ratio of either G M1 or G M2 , the molar ratio between G M2 activator and the respective ganglioside in the complex was found to be approximately 1:1. However, when the activator protein was incubated with 50 molar excess of either G M1 or G M2 , the molar ratio between the ganglioside and G M2 activator in the complex was found to be about 50:1 in both cases. When the bindings between G M2 activator and a 50-fold molar excess of G M3 was examined, we found that the ratio of G M3 to the activator protein was about 80:1. It is well documented that in an aqueous medium G M3 exists as vesicles that are larger than micelles (31). Therefore, it is not surprising to find that the ratio of G M3 /activator protein to be larger than that of G M2 /activator protein or G M1 /activator protein. As also shown in Table II, the association of saposin B to G M2 was very similar to that of G M2 activator protein. Thus, the interactions between the activator and the gangliosides detected in aqueous medium are similar for saposin B and G M2 activator. Whereas, the bindings on TLC overlay showed that saposin B bound to all glycosphingolipids, and G M2 activator bound preferentially to the anionic glycosphingolipids.
The Bindings of G M2 Activator to the Mixture of G M1 and G M2 or G M2 and G M3 -We have reported that G M2 activator was able to recognize the branched trisaccharide epitope of G M2 (8,9). We, therefore, examined whether G M2 activator can specifically bind only to G M2 when G M2 was mixed with G M1 or G M3 . G M2 was first mixed with G M1 or G M3 in chloroform:methanol (2/1, v/v), dried, and redispersed in an aqueous buffer solution. The aqueous ganglioside mixture was then incubated with G M2 activator and subjected to Sephacryl S-200 gel filtration as described under "Experimental Procedures." The complexes were isolated, and the amounts of G M2 , G M1 , and the activator protein (or G M2 , G M3 , and the activator protein) were determined. As shown in Table III, G M2 activator did not appear to bind preferentially to G M2 to form the activator protein⅐G M2 complex in 1:1 ratio. Rather, it associated with the mixture of gangliosides in the proportion similar to that in the original ganglioside mixture. For example, when G M2 activator was incubated with a mixture containing an equimolar ratio of G M1 and G M2 , the molar ratio of G M2 activator, G M1 , and G M2 in the complex was found to be close to 1:1:1. However, when G M2 activator was incubated with a mixture containing 25-fold excess of G M1 and G M2 , the detected ratio of the activator protein to G M1 and G M2 in the complex was 1:17.3:17.0. No preferential extraction of G M2 from the two ganglioside mixtures was observed. A similar result was obtained from the incubation of G M2 activator with a mixture of G M2 and G M3 . These results indicate that the composition of the complexes formed under the micellar form of ganglioside was determined by the preexisting status of the ganglioside micelles. DISCUSSION Among the five activator proteins that stimulate the enzymatic hydrolysis of glycosphingolipids, saposin B and G M2 activator have been shown to interact and affect the glycosphingolipid substrates (29,30). Several methods have been used to demonstrate the complex formation between the activator proteins and glycosphingolipids, and in some studies the molar ratios between the protein and the lipid were also determined. For example, Fischer and Jatzkewitz (32) studied the complex formation between saposin B and sulfatide using electrophoresis and reported that the ratio of these two components in the complex was 1:1. Also using electrophoresis, Wenger and Inui (33) reported the ratio of the two compounds in the saposin B⅐G M1 and saposin B⅐sulfatide complexes to be 1:4 and 1:2.6, respectively. Vogel et al. (34) studied the binding of saposin B to the individual gangliosides, such as G M1 , G M2 , G M3 , and G D1a , as well as sulfatide by centrifugation and determined the molar ratios between saposin B and each of these gangliosides in the protein-lipid complexes to be almost 1:1. For G M2 activator, Conzelmann et al. (4,10) concluded from their studies using ultracentrifugation, isoelectric focusing, and electrophoresis that G M2 activator can form the activator protein⅐G M2 complex in 1:1 ratio. These experiments were carried out under the conditions required for the specific methodology used (for example, high sucrose density for ultracentrifugation and high pH for electrophoresis). Using TLC overlay, we have shown clearly that the high ionic strength or high pH of the buffer FIG. 6. TLC analysis showing the conversion of G M2 into G M3 by ␤-hexosaminidase A in the G M2 ganglioside⅐G M2 activator complex. G M2 activator (1.34 nmol) was preincubated with 50 molar excess of G M2 , and the entire mixture was subjected to gel filtration through a Sephacryl S-200 column. The complex was isolated and incubated with ␤-hexosaminidase A and analyzed by TLC as described in the text. Lane l, the complex was incubated without ␤-hexosaminidase A; lane 2, the complex was incubated with ␤-hexosaminidase A; lane 3, the standard G M2 was incubated with ␤-hexosaminidase A but without the activator protein; lane 4, G M2 standard; and lane 5, G M3 standard.  solution inhibited the interactions between G M2 activator and the glycosphingolipid substrates. Therefore, we chose to analyze the complex formation between G M2 activator and gangliosides using 25 mM ammonium acetate buffer, pH 4.0, which is optimal for the enzymatic hydrolysis of G M2 in the presence of G M2 activator. By TLC overlay, G M2 activator was found to bind to various anionic glycosphingolipids without showing preference to any particular sugar chain. Thus, G M2 activator does not behave like lectins, which display the recognition of specific saccharide structure. The involvement of an anionic residue of a glycosphingolipid in the complex formation with an activator protein has been suggested. We have reported that the carboxylic function of the NeuAc in G M2 was important for the action of G M2 activator (16). Also, Mitsuyama et al. (35) have reported the binding of saposin B to the affinity column packed with the immobilized sulfatide or its derivatives as ligands. In the present studies using TLC overlay, we have clearly demonstrated that the anionic group in glycosphingolipids is vital for the complex formation with G M2 activator but not with saposin B. While the bindings of G M2 activator to gangliosides and sulfatides are greatly affected by the assay conditions, such as the pH and the ionic strength of the buffer solutions (Figs. 1, 3, and 4), no such effects were found for saposin B. These results, again, support the importance of the negative charge in a glycosphingolipid to form the glycolipid⅐G M2 activator complex.
Wynn (36) proposed the triple binding domain theory of a glycosphingolipid to saposin B based on the conformational studies of the glycosphingolipids. He predicted that there are three possible interactions between a glycosphingolipid and the protein: (a) the hydrophobic interaction of the hydrocarbon chains of the ceramide moiety and a complementary hydrophobic domain in the protein molecule; (b) the electrostatic interaction between sialic acid or sulfate group and a positively charged group of the protein; and (c) the hydrophilic interaction between a hydroxyl group in a sugar moiety and a complementary plane of the protein. Wynn (36) also pointed out that the glycolipid which has at least two of these structural features will strongly bind to saposin B. Our results on the binding behavior of both saposin B and G M2 activator toward glycosphingolipids agree well with this model, since we have shown that G M2 activator was not able to distinguish the saccharide backbone, the number of sugar residues, and the position or the nature of sialic acid (Table I). It is evident that G M2 activator is not specific to bind only G M2 . As both saposin B and G M2 activator were shown to be able to transport glycosphingolipids from the donor to the acceptor liposomes (10), G M2 activator may have a specific role in vivo to transport the acidic glycosphingolipids.
Our results on the complex formation between G M2 activator and the micellar forms of gangliosides agree well with the studies of Cantu et al. (37). They studied the micelle formation in mixed gangliosides using light scattering and neutron scattering and reported that when G M2 and G T1b were mixed in different molar ratios in aqueous solution, the two gangliosides formed a single family of mixed micelles rather than that of two families of unmixed micelles, and the ratio of each ganglioside in the mixed micelles depended on the molar concentration of each ganglioside (37).
In contrast to previous reports (4, 10), we were not able to explain the mode of action of G M2 activator based on our studies on the complex formation between this activator protein and glycosphingolipids, especially G M2 . The fact that G M2 activator interacts with a wide variety of anionic glycosphingolipids indicates that this activator protein may have functions other than assisting the enzymatic hydrolysis of G M2 .