Self-recognition of N-linked glycans with multivalent GlcNAc, determined as ceramide mimetic conjugate

Abstract

Aminoceramide mimetic was synthesized and conjugated to N-linked oligosaccharides having multivalent GlcNAc by reductive amination. Ceramide mimetic conjugates with “complex-type” glycan having five or six GlcNAc termini (termed Os Fr. B-Cer) were purified, analyzed by thin-layer chromatography (TLC), and finally characterized by MS/MS analysis through liquid chromatography/mass spectrometry. Binding of Os Fr. B-Cer placed on solid phase polystyrene surface with [³H]cholesterol-labeled liposomes containing Os Fr. B-Cer, or containing various glyco- sphingolipids (GSLs) was determined. The binding of Os Fr. B-Cer liposomes to Os Fr. B-Cer coated plate was significantly higher than binding of GM3 liposomes. Other GSL liposomes showed no binding. Thus, self-recognition of Os Fr. B-Cer was clearly demonstrated using ceramide mimetic conjugates.

aminoceramide, carbohydrate-to-carbohydrate interaction, glycosphingolipid, homotypic interaction, oligosaccharide

Introduction

One of the major functional roles of glycosylation is based on recognition of glycosyl residues expressed at the cell surface as glycosphingolipids (GSLs) or N-linked or O-linked glycan of glycoproteins, by a number of carbohydrate-binding proteins (endogenous lectins): galectins (Leffler 2004), selectins (Varki 1994; Magnani 2004), siglecs (Crocker and Varki 2001), and many glycosaminoglycan-binding proteins (Esko 1999). A different, novel mechanism of carbohydrate recognition is carbohydrate-to-carbohydrate interaction (CCI), as a basis of cell-to-cell adhesion. This process was discovered ∼20 years ago (Eggens et al. 1989; Misevic and Burger 1993), but has received little attention.

Two types of CCI were initially found. One is based on self-recognition of Le^x by Le^x expressed in GSL or glycoprotein of embryonic stem cells or teratocarcinoma, leading to self-adhesion (autoaggregation) (Eggens et al. 1989; Kojima et al. 1994). The other system is self-recognition of proteoglycans expressed on sponge cells, causing species-specific aggregation of sponge cells (Misevic and Burger 1990; Misevic and Burger 1993). Recent studies indicate that the glycan involved in self-recognition in a particular sponge species, Microciona prolifera, has multiple fragments of 3-O-sulfated GlcNAcβ3 l-Fucα-O or 4,6-pyruvated Galβ4GlcNAcβ3 l-Fucα (Spillmann et al. 1993; Spillmann et al. 1995; Haseley et al. 2001; Carvalho de Souza et al. 2005). Other sponge species may have different glycan structures with different self-recognition specificity (Bucior and Burger 2004a; Bucior et al. 2004b). Thus, self-recognition of glycosyl epitopes leads to homotypic interaction of cells causing autoaggregation. On the other hand, glycosyl epitopes expressed in GSLs on one type of cell may interact with a different glycosyl structure expressed on a different type of cell, e.g., based on binding of GM3 to LacCer (Kojima and Hakomori 1991), to Gg3Cer (Kojima and Hakomori 1989).

During the past several years, CCI has been further elaborated and quantitatively determined based on the advanced technology using novel synthetic approach. Examples are molecular force microscopy applied to Le^x-to-Le^x interaction (Tromas et al. 2001), and surface plasmon resonance (SPR) spectros- copy applied to interaction of Gg3 with GM3 (Matsuura et al. 2000) or self-interaction of sponge cell oligosaccharides (Os) (Haseley et al. 2001). Self-interaction of specific Os was also observed based on autoaggregation of Le^x Os affixed on gold nanoparticles in the presence of Ca²⁺ (de la Fuente et al. 2001), and autoaggregation of sponge cell Os affixed on gold nanoparticles, also with Ca²⁺ (Carvalho de Souza et al. 2005).

However, studies on CCI are still limited to a relatively small number of structures. Recently, we found that GM3 interacts with novel N-linked glycans having multivalent GlcNAc (termed “Os Fr. B”) (Yoon, Nakayama, Takahashi, et al. 2006). In this study, binding of GM3 liposome to Os Fr. B was observed when Os Fr. B was conjugated to phosphatidylethanolamine (PE). Os conjugation to PE was performed by a previously published method (Stoll and Feizi 1997). However, binding efficiency of such Os-PE conjugate was not as high as that of GSL-liposome binding to GSL-coated plate (see Discussion). In the present study, therefore, Os Fr. B or other Os were conjugated to aminoceramide mimetics, and binding of such Os-ceramide conjugate, incorporated in ³H-labeled liposomes, to GSL-coated or Os-ceramide-coated polystyrene plates was determined. Os-ceramide conjugate bound much more strongly than PE-conjugate. Os Fr. B-ceramide (Os Fr. B-Cer) incorporated into liposome bound strongly to Os Fr. B-Cer per se coated on plate, indicating that Os Fr. B is capable of self-binding.

Results

Synthesis of aminoceramide mimetics

This synthesis was performed starting from commercially available 2-tetradecylhexadecanoic acid (compound 1 in Figure 1) through conversion of its carboxyl group to hydroxyl (compound 2) (Hasegawa et al. 1996). Compound 2 was converted to intermediate compound 3 by novel Mitsunobu reaction (Mitsunobu 1981). Compound 3 was converted to aminoceramide (compound 4) by reaction with ethylenediamine. Properties of each reaction product, particularly compounds 3 and 4, were characterized in detail by MS and NMR, as described in Materials and methods. Compound 3 showed main mass signal with m/z 568.68, and characteristic ¹H NMR pattern. Compound 4 showed main mass signal with m/z 438.7, identified as [M+H]⁺ of aminoceramide (Figure 2). Very minor mass signals m/z 719.8 and 778.8 were not assigned properly.

Fig. 1.

Reaction scheme for preparation of aminoceramide. Aminoceramide was synthesized as described in Materials and methods. Compound 1, 2-tetradecylhexadecanoic acid (C₃₀H₆₁O₂, Mw 453), Compound 2, 2-tetradecylhexadecanol (C₃₀H₆₂O, Mw 438), Compound 3, intermediate (C₃₈H₆₅O₂ N, Mw 567), Compound 4, aminoceramide (C₃₀H₆₃ N, Mw 437).

Os Fr. B and its conjugate with NH₂Cer

Os Fr. B (Figure 3A) reacted with aminoceramide by reductive amination (Figure 3B) yielded a component bound to C18 column, eluted with C/M/W 2:1:0.1. This component (spot 2, lane b, Figure 4A) was reactive with both orcinol/sulfuric acid and primulin spray, and showed faster TLC migration than original Os Fr. B (spot 1, lane a, Figure 4A), which was reactive with orcinol/sulfuric acid but not with primulin spray. The Os Fr. B-Cer conjugate (spot 2, with dual reactivity to orcinol and primulin) was further purified by solvent partitioning. Insoluble fraction in I/H/W 55:40:5 contained the Os Fr. B-Cer but not aminoceramide. The insoluble fraction contained two major bands corresponding to Os Fr. B1-Cer and Os Fr. B2-Cer, both reactive with orcinol and with primulin (lane e, Figure 4B).

Fig. 2.

Positive-ion ESI-MS spectra of aminoceramide, and MS/MS data for each major mass. (Panel A) Aminoceramide. Note the presence of one major ions, with mass number 438.7, corresponding (M+H)⁺ major ions with (C₃₀H₆₃ N). (Panel B) MS/MS spectrum of m/z 438.7 corresponding to aminoceramide, (C₃₀H₆₃ N), of which product ions of m/z based on loss of one CH₂⁻.

Fig. 3.

Structures of major Os Fr. B and its coupling reaction with aminoceramide. (Panel A) Os Fr.B is a mixture of structures with and without Galβ1-4 substitution to a defined GlcNAc as shown. Structural assignment was based on HPLC mapping method (“GALAXY”), and some were confirmed by MALDI-TOF-MS analysis, as described in Material and methods. (Panel B) Os Fr. B was conjugated with aminoceramide (NH₂Cer) in the present of sodium cyanoborohydride as described in Material and methods. Aldehyde group in reducing terminal of N-GlcNAc of Os Fr. B reacts with amino termini of aminoceramide in the presence of sodium cyanoborohydride, producing the Os Fr. B-Cer conjugate.

Fig. 4.

TLC patterns of Os Fr. B-Cer derivatives. (Panel A) Preliminary separation of Os Fr. B-Cer conjugates using C₁₈-cartridge. Coupling reaction mixture was subjected to C₁₈-cartridge as described in Material and methods. (Panel B) Purification of Os Fr. B-Cer from C₁₈-cartridge-bound fraction. C₁₈-cartridge-bound fraction was dried and dissolved in I/H/W 55:40:5, divided by centrifugation, and dried under N₂ stream. (lane a) Os Fr. B; (lane b), C₁₈-cartridge-bound fraction; (lane c), synthesized control NH₂Cer; (lane d), soluble fraction in I/H/W 55:40:5 from compound in lane b; (lane e), insoluble fraction in I/H/W 55:40:5 from compound in lane b. Upper panels were detected by primulin spray for lipids, and lower panels were detected by orcinol/sulfuric acid reaction for carbohydrate. TLC solvent, C/M/W 50:55:18.

Structural characterization of Os Fr. B-Cer conjugate by mass spectrometry

The Os Fr. B-Cer conjugate thus prepared was analyzed by electrospray mass spectrometry (ES-MS). Two mass signals with m/z 1278.8 and 1359.8 were identified as [M+2H]²⁺ of Os Fr. B1-Cer and of Os Fr. B2-Cer, respectively (Figure 5A). MS/MS analysis of m/z 1359.8 and 1278.8 gave the patterns shown in Figure 5B and C, respectively. The major peak with m/z 1258 is [M-HexNAc+2H]²⁺ from m/z 1359, corresponding to Os Fr. B2-Cer (Figure 5B). The peak with m/z 1177 is [M-HexNAc+2H]²⁺ from m/z 1279, corresponding to Os Fr. B1-Cer (Figure 5C). The data indicate that these Os Fr. B-Cer conjugates have correct structure. LC-MASS of Os Fr. B-Cer and the ions yielded by MS/MS are shown in Figure 5D.

Fig. 5.

Positive-ion ESI-MS spectra of aminoceramide conjugates of Os Fr. B, and MS-MS data for each major mass. (Panel A) Aminoceramide conjugates of Os Fr. B. Note the presence of two major ions, with mass number 1278.8 and 1359.8, corresponding respectively to doubly charged (M+2H)²⁺ major ions with (8 GlcNAc, 3 Man, NHCer) and (1 Gal, 8 GlcNAc, 3 Man, NHCer). Other minor ions are well assigned by elimination of GlcNAc or Gal from the two major ions. (Panel B) MS/MS spectrum of m/z 1359.8 corresponding to Os Fr. B2-Cer (product ion of m/z 1258.2 based on loss of one HexNAc from m/z 1359.8). (Panel C) MS/MS spectrum of m/z 1278.8 corresponding to Os Fr. B1-Cer (product ion of m/z 1177.2 based on loss of one HexNAc from m/z 1278.8). (Panel D) Possible assigned structures of Os Fr. B-Cer conjugates corresponding to expected structure, based on results of MS/MS.

Binding of Os Fr. B-Cer to Os Fr. B-Cer, and to GM3

Os Fr. B-Cer was coated on polystyrene multiwell plates, to which binding of PC/[³H]cholesterol liposome incorporating Os Fr. B-Cer or other GSLs was tested as described in Materials and methods. Results are shown in Figure 6. Os Fr. B-Cer-liposomes bound strongly to Os Fr. B-Cer coated plates. GM3-liposomes also showed clear binding, whereas GalCer-, Forssman-, and LacCer-liposomes showed no binding.

Fig. 6.

Binding of ³H-labeled liposome containing Os Fr. B-Cer or other GSLs to various Os Fr. B-Cer derivatives coated and affixed on multiwell polystyrene plates. Various quantities (0–0.8 μg/well) of various Os Fr. B-Cer derivatives as shown on abscissa were dried at 37°C, blocked with 1% BSA for 1 h, and washed. Each well was added with 100 μL of ³H-labeled liposome containing Os Fr. B-Cer, GM3, LacCer, Forssman antigen, or GalCer, incubated for 16 h, and degree of binding was determined, as described in Materials and methods. Clear GM3 liposome binding was observed for Os Fr. B-Cer. Binding of Os Fr. B-Cer was strong. No other GSL-liposomes binding were observed for Os Fr. B-Cer. Data shown are typical results from a single triplicate experiment. Similar results were obtained in two other triplicate experiments. Bars indicate standard deviation.

Discussion

Studies on CCI in general are still limited to a relatively small number of carbohydrate structures, and our studies have been based on GSL-liposome binding to GSL-coated plate (Eggens et al. 1989; Kojima and Hakomori 1991), or autoaggregation of GSL-coated polystyrene beads (Kojima et al. 1994). Studies from other labs have been greatly progressed by various chemical and physical methods. A few examples are: (i) self-aggregation of glycosyl epitopes indicated by autoaggregation of gold sol nanospheres in the presence of Ca²⁺ (Rojo et al. 2004; Carvalho de Souza et al. 2005; de la Fuente et al. 2005) [for review see (de la Fuente and Penades 2004)], (ii) determination of atomic force between glycosyl residues at cantilever tip and those affixed on self-assembled gold film (Tromas et al. 2001), (iii) interaction of GSL, as Langmuir monolayer affixed to hydrophobized gold film on glass, with specific glycosyl epitope bound to polystyrene, and recording by SPR (Matsuura et al. 2000), (iv) interaction between the same components as in (iii), determined as change of π-A (Δπ) (Matsuura et al. 1998), or interaction of GSL Langmuir monolayer with micellar form of lyso-GSL determined as Δπ (Santacroce and Basu 2003), and (v) interaction of two osmotically controlled vesicles containing the same or different GSLs held in micropipets by aspiration; intensity of interaction is determined by contact angle θ_c (Gourier et al. 2005).

Interaction of glycosyl epitopes expressed on GSL has been extended to GSL interaction with N-linked glycan, since there is a high probability of GSL interaction through their clusters with adjacent glycans in membrane proteins. Systematic studies revealed that only complex-type N-linked glycans having multiple GlcNAc termini, but not complex types with complete glycosylation (sialyl-Gal linked to GlcNAc), and not high-mannose type, interacts with GM3 (Yoon, Nakayama, Takahashi, et al. 2006). In this study, N-linked glycans released by hydrazinolysis and purified by multistep procedure were linked to PE (Feizi et al. 1994; Stoll and Feizi 1997). Yield of PE conjugation of highly complex Os is low, and the reaction takes a long time (72–96 h). In addition, PE conjugate is difficult to incorporate in liposome. Therefore, interaction between the same N-linked glycan Os, or pairs of N-linked glycan Os, is impossible to observe.

To overcome this problem, we used aminoceramide mimetics. In general, lipids with high level of saturated fatty acid, in contrast to lipids with high level of unsaturated fatty acid (e.g., oleic and linoleic acid), may display higher degree of side-by-side (cis) interaction to form clusters, which are separable as detergent-resistant, low-density membrane fraction (Brown and Rose 1992; Brown and London 1997). Naturally occurring ceramide containing sphingosine has N-acyl group and hydroxyl group, and therefore has a higher chance of hydrogen bond formation between sphingollipids than between phospholipids. This is another cause of GSL clustering (Pascher 1976; Hakomori et al. 1998).

Ceramide mimetics used in this study consist of saturated aliphatic chain within PC/cholesterol membrane, and their Os conjugates may form clusters. Os Fr. B (consisting of Os Fr.B1 and Os Fr.B2) was prepared and conjugated with aminoceramide by reductive amination, followed by extensive purification. Os Fr. B-Cer thus prepared was capable of getting incorporated in liposome, and it was possible to determine its interaction with the same Os Fr. B-Cer coated on plate. Remarkably, the liposomes showed strong self-interaction. They showed moderate interaction with GM3 liposome, but no interaction with other GSL-liposomes.

Self-recognition of glycosyl epitopes provides a basis of homotypic cell-to-cell binding, as observed in adhesion of embryonal stem cells and F9 cells (Eggens et al. 1989; Kojima et al. 1994), and in sponge cell autoaggregation (Spillmann et al. 1993, 1995; Haseley et al. 2001; Bucior et al. 2004b; Carvalho de Souza et al. 2005), as described in introduction. Self-recognition of Os Fr. B may also provide an important basis of either homotypic interaction of certain types of cells, since N-linked glycans with GlcNAc termini were claimed to be involved in interaction of sperm cells with Sertoli cells during spermatogenesis (Akama et al. 2002). Interestingly, the mAb J1 was reported to define a stage-specific glycosyl epitope associated with differentiation of male germ cells (Fenderson et al. 1984). An epitope of J1 was subsequently determined as GlcNAc termini of a glycoprotein as well as GSL, although exact variation of structure was not clearly identified (Symington et al. 1984). A possibility is open that self-recognition of N-linked glycans having multiple GlcNAc termini, similar to Os Fr. B, may mediate a number of biological processes, including spermatogenesis.

Studies of interaction of N-linked glycans are limited to complex types with exposed GlcNAc termini (Yoon, Nakayama, Hikita, et al. 2006; Yoon, Nakayama, Takahashi, et al. 2006). GM3 interaction has so far been observed clearly with (i) complex types with 5-6 GlcNAc termini, which includes bisecting GlcNAcβ4 Man; and (ii) complex types with 3-5 GlcNAc termini, derived by desialylation and degalactosylation of tri- or tetra-antennary structure. However, GM3 does not interact, or interacts weakly, with N-linked glycan having 2 GlcNAc termini without bisecting GlcNAcβ4 Man, as observed in “Os Fr.1”, which has bi-antennary structure with GlcNAcβ2 Manα6 and GlcNAcβ4 Manα3, both linked to Manβ4GlcNAcβ4GlcNAc (Yoon, Nakayama, Takahashi, et al. 2006). Further study is needed to clarify the functional role of bisecting GlcNAc in interaction with GM3, as well as self-interaction of N-linked glycans. These previous studies clearly indicate that substitution of GlcNAc by Gal blocks the interaction with GM3. However, the effect of Gal blocking of GlcNAc termini on self-recognition of Os Fr. B remains to be studied. It is also important to note that high-mannose type does not interact with GM3 (Nakayama K, Yoon S, Hakomori S, unpublished data).

Functional roles of carbohydrates are to (i) modify conformational structure and function of proteins and (ii) mediate adhesion or other processes through carbohydrate-binding proteins. In either (i) or (ii), carbohydrate in GSL or in glycoprotein plays a major role in defining cellular function. Results of the present study indicate a new possibility for the role of carbohydrate-to-carbohydrate interaction through N-linked glycans in defining cell adhesion, signal transduction, and many other cellular phenotypes.

Materials and methods

Materials and reagents

All reagents were from Sigma Chemical Co. (St. Louis, MO), except for the following items. GM3 (having NeuAc) was prepared from dog erythrocytes (Yamakawa 1991). Forssman GSL antigen was prepared from sheep or goat erythrocytes (Papirmeister and Mallette 1955). Gg3 was prepared from guinea pig erythrocytes (Seyama and Yamakawa 1974). GalCer and LacCer were from Matreya (Pleasant Gap, PA). [³H]cholesterol (0.25 mCi/0.25 mL) was from PerkinElmer. Dialysis membrane (MWCO 12000-14000) was from Spectrum Labs (Rancho Dominguez, CA). Solvents (HPLC grade) were from Fisher. Distilled water was from a Milli-Q purification system (Millipore, Bedford, MA).

Oligosaccharides, Os Fr. B1 and Os Fr. B2, from ovalbumin

N-linked glycans from ovalbumin were released by hydrazinolysis in argon atmosphere, followed by removal of hydrazine, N-acetylation, and cellulose column chromatography to separate N-acetylated Os fraction (Mizuochi 1993; Shimizu et al. 2001). The yield of cellulose column-bound Os fraction was 111 mg (11.1% yield) from 1 g ovalbumin. The fraction was further separated into ConA-binding (mainly high-mannose type Os) versus ConA-nonbinding (including hybrid-type and complex-type Os with multivalent GlcNAc) components as described previously (Ohyama et al. 1985; Merkle and Cummings 1987). The yield of ConA-nonbinding fraction was 11.0 mg (9.9% yield) from 1 g ovalbumin (Yoon, Nakayama, Takahashi, 2006).

ConA-nonbinding Os was further separated into various Os components by HPLC through normal-phase Microsorb 100 NH₂ column (Varian, Lake Forest, CA; 0.46 × 25 cm) using a Varian ProStar chromatography system. Separations were performed at room temperature. Mobile phase consisted of Sol A (90%, v/v, 0.2 M acetic acid/triethylamine, pH 7.2: acetonitrile = 3:7) and Sol B (10%, v/v, 0.2 M acetic acid/triethylamine, pH 7.2: acetonitrile = 7:3) to start, followed by a linear gradient from 90 to 80% over 50 min. During the next 10 min, the percentage of Sol B was maintained at 100%. The flow rate was 1 mL/min, fraction size was 1 mL, and the effluent was monitored using a Waters 484 tuneable absorbance UV detector at λ = 215 nm. The major homogeneous Os components (Os Fr. B) were thus separated by HPLC. These Os components were subjected to structural analysis (see below).

Os structures were identified as 2-aminopyridine derivatives by HPLC mapping method as described previously (Tomiya et al. 1988; Takahashi et al. 1993; Takahashi et al. 1995). The PA-Os were identified by comparison with HPLC data of approximately 500 reference PA-Os in a home-made web application, GALAXY (http://www.glycoanalysis.info/) (Takahashi and Kato 2003).

Synthesis and properties of intermediate compound 3 and end-product of aminoceramide (NH₂Cer) mimetics

Synthesis was initiated by conversion of 2-tetradecyl- hexadecanoic acid (compound 1 in Figure 1), purchased from Wako Pure Chemical Industries, Ltd. (Osaka, Japan), to 2-tetradecylhexadecanol (compound 2), by the procedure described previously (Hasegawa et al. 1996). Phthalimide was bound to hydroxyl group of 2-tetradecylhexadecanol by dehydration through Mitsunobu reaction (Mitsunobu 1981) as a key step, to yield intermediate compound 3, which is novel. To a solution of 2-tetradecylhexadecanol (313 mg, 710 μmol), phthalimide (105 mg, 710 μmol) and triphenylphosphine (186 mg, 710 μmol) in tetrahydrofuran (THF, 1.71 mL) was added a solution of diethyl azodicarboxylate (DEAD, 110 μL, 710 μmol) in THF (300 μL) at room temperature. After stirring for 96 h under reflux, the reaction mixture was concentrated and extracted with diethyl ether. The organic layer was washed with brine, dried over anhydrous sodium sulfate, and concentrated to a syrup, which was chromatographed on a column of silica gel (n-hexane-ethyl acetate 300:1) to give compound 3 (279 mg, 69.2%). The structure was analyzed by MALDI-TOF MS and NMR. The main mass signal with m/z 568.68 was identified as [M+H]⁺ of the compound. ¹H NMR (CDCl₃) at 500 MHz: δ 0.86–0.89 (t, 6 H, CH₃), 1.20–1.36 (s, 52 H, CH₂), 1.88 (m, 1 H, H-2), 3.56–3.58 (dd, 2 H, H-1), and 7.70–7.85 (m, 4 H, aromatic H).

Compound 3 was converted to aminoceramide (compound 4) by reaction with ethylenediamine. To a solution of compound 3 (279 mg, 491 μmol) in n-butanol (10 mL) was added ethylenediamine (2 mL, 18 μmol), and the mixture was stirred for 25 h at 90°C. After concentration with addition of toluene and ethanol, the residue was chromatographed on a column of Sephadex LH-20 (chloroform-methanol 200:1) to give aminoceramide (190 mg, 88.4%).

Conjugation of NH₂Cer to Os, and purification of product

Os Fr. B (Mw 2211) was conjugated to aminoceramide as follows. Briefly, 1 mg (0.5 μmol) aliquot of Os Fr. B in vial was completely dried in a desiccator overnight under vacuum, and dissolved in 60 μL distilled water. After addition of 427 μL of methanol and 113 μL (1.3 μmol) of NH₂Cer (5 mg/mL, v/v, in methanol, pH 6.2), the atmosphere of the vial was exchanged with argon gas, the vial was tightly sealed with teflon-lined cap, and heated at 100°C for 30 min. The vial was cool downed, 5 μL (12 μmol) of sodium cyanoborohydride (150.8 mg/mL, v/v, in distilled water) was added, and incubated at 90°C for 16 h. The reaction mixture as above was dried under N₂, dissolved in 1.5 mL methanol:water 6:4 and applied to a small pre-packed C₁₈-alkylated silica gel (“C₁₈-cartridge”) (Varian) equilibrated with methanol:water 6:4. The column was washed with 10-column volumes methanol:water 6:4 to eliminate all water-soluble contaminants. The adsorbed Os Fr. B-Cer conjugate and nonreacted NH₂Cer were eluted from C₁₈ cartridge with chloroform/methanol/water (C/M/W) 2:1:0.1, dried under N₂ stream, and dissolved in I/H/W 55:40:5. After centrifugation at 13,000 rpm for 5 min, the pellet dissolved in C/M/W 25:25:8 and monitored by high performance TLC analysis, developed with C/M/W 50:55:18. Spots were revealed by primulin spray (Skipski 1975) and orcinol/sulfuric acid reaction. The dried residue was dissolved in a defined volume of C/M/W 25:25:8. The concentration of Os Fr. B-Cer conjugate was measured based on primulin (0.001% primulin/aceton:H₂O = 4:1) and orcinol/sulfuric acid reactivity (0.2% orcinol/H₂SO₄:H₂O = 1:9), by TLC with Scion image analysis, and aliquots of suitable quantity were used for binding assays with GM3 and other GSLs.

Characterization of structure of N-linked Os Fr. B conjugates with NH₂Cer by ESI mass spectrometry

Structures of Os Fr. B-Cer conjugates were confirmed by mass spectrometry using an ion trap mass spectrometer Esquire LC (Bruker Daltonics, Billerica, MA, USA) with electrospray ionization source. To confirm the presence of structures of Os after NH₂Cer-coupling reaction, 25 μg of Os Fr. B-Cer conjugate dissolved in C/M/W 25:25:8 was diluted 10-fold in methanol/1% formic acid/10% ammonium acetate. This solution was directly infused into the ion source at a flow rate of 1 μL/min. Spectra were collected in both positive and negative ionization mode. The fragmentation spectra of the analytes were collected with an isolation width of 4 amu, fragmentation amplitude set to 1 V and SmartFrag On (amplitude automatically varied to 30–200% of the set fragmentation amplitude of 1 V).

Preparation of PC/cholesterol liposome incorporating Os Fr. B-Cer or naturally-occurring GSLs, and their binding to Os Fr. B-Cer coated on polystyrene beads

Various quantities (0.2–0.8 μg per well, calculated based on equimolar quantities) of Os Fr. B-Cer conjugate in 50% aqueous ethanol were added to each well of 96-well flat-bottom polystyrene plates (Costar #9017, Corning Inc., Acton, MA). The coated plates were dried for 5 h at 37°C, 100 μL 1% BSA was added to each well and incubated 1 h to block nonspecific binding, and wells were washed with TBS solution (see below). ³H-labeled liposomes containing GSL or Os Fr. B-Cer conjugate were prepared as described previously (Eggens et al. 1989; Kojima and Hakomori 1989). Briefly, 50 μg 1,2 dimyristoyl-sn-glycero-3-PC, 30 μg nonlabeled cholesterol, 2 μL ³H-labeled cholesterol (0.25 mCi per 0.25 mL), and 10 μg Os Fr. B-Cer conjugate or respective GSL (GM3, GalCer, Forssman, LacCer) were dissolved in 200 μL C/M 2:1, and the solution was evaporated to dryness in a rotary evaporator. The dried residue was mixed with 2 mL TBS solution (10 mM Tris-HCl (pH 8.0)/ 0.9 mM CaCl₂/0.5 mM MgSO₄/0.1 mM MnCl₂), vigorously mixed with a Vortex mixer, and sonicated (Bransonic, model 5510R-MT, Branson Ultrasonics Corp., Danbury, CT) at ∼20°C for 2 h. Temperature increase of the sonication bath was prevented by occasional addition of ice. In all experiments, freshly prepared liposomes in TBS solution were used. To each well coated with Os Fr. B-Cer conjugate as above, 100 μL liposome preparation was added, the plate covered by lid was wrapped with polyvinyl film, and incubated for 16 h at room temp with shaking (Red Rocker PR50). Wells were washed three times with TBS. For effective washing, 200 μL TBS was added to each well, the tip of a thin pipette was placed at the edge, and washing medium was repeatedly sucked out. The last washing was tested for radioactivity. Bound liposomes were extracted twice with I/H/W 55:25:20, and radioactivity of the extract was counted.

Funding

National Institute of Health/National Institute of General Medical Science (R01 GM070593 to S.H.); Ministry of Education, Culture, Sports, Science, and Technology (MEXT) of Japan (Grant-in-Aid for Scientific Research, No. 17101007 to M.K.).

Conflict of interest statement

None declared.

Abbreviations

 
• CCI

  carbohydrate-to-carbohydrate interaction

 
• GalCer

  β-galactosylceramide

 
• Gg3

  GalNAcβ4Galβ4Glcβ1Cer

 
• GM3

  NeuAcα3Galβ4Glcβ1Cer

 
• GSL

  glycosphingolipid

 
• LacCer

  lactosylceramide (Galβ4Glcβ1Cer)

 
• Os

  oligosaccharide

 
• PE

  phosphatidylethanolamine

 
• TLC

  thin-layer chromatography.

© Oxford University Press 2007